nohup: ignoring input Namespace(save_dir='/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg', self_attn_layer_to_quant='4 1 2 8 23 22 25 5 24 7 26 6 20 12 19 17 21 11 10 9 18', mlp_layer_to_quant='27 16 19', model_id='/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen/Qwen2.5-7B', cuda_id=6) `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2239781:2240030 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2239781:2240030 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2239781:2240030 [0] NCCL INFO Using network IB n136-128-154:2239782:2240031 [1] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2239782:2240031 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2239782:2240031 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2239782:2240031 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2239782:2240031 [1] NCCL INFO Using network IB n136-128-154:2239782:2240031 [1] NCCL INFO ncclCommInitRankConfig comm 0x10b4d040 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xc1a41f07abad06fa - Init START n136-128-154:2239781:2240030 [0] NCCL INFO ncclCommInitRankConfig comm 0x102eb980 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xc1a41f07abad06fa - Init START n136-128-154:2239781:2240030 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2239782:2240031 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2239781:2240030 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2239781:2240030 [0] NCCL INFO Retrieving state for IB n136-128-154:2239781:2240030 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2239781:2240030 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2239781:2240030 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2239781:2240030 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2239782:2240031 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2239782:2240031 [1] NCCL INFO Retrieving state for IB n136-128-154:2239782:2240031 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2239782:2240031 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2239781:2240030 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2239782:2240031 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2239782:2240031 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2239782:2240031 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2239782:2240031 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2239782:2240031 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2239781:2240030 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2239781:2240030 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2239782:2240031 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2239782:2240031 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2239781:2240030 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2239781:2240030 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2239782:2240031 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2239782:2240031 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2239782:2240031 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2239781:2240030 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2239782:2240031 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2239781:2240030 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2239782:2240031 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2239781:2240030 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2239782:2240031 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2239781:2240030 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2239782:2240031 [1] NCCL INFO ========================================== n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2239782:2240031 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2239781:2240030 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2239781:2240030 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2239781:2240030 [0] NCCL INFO ========================================== n136-128-154:2239782:2240031 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2239781:2240030 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2239782:2240031 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2239781:2240030 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2239781:2240030 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2239781:2240030 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2239782:2240031 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2239781:2240030 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2239781:2240030 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2239781:2240030 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239782:2240031 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2239781:2240030 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239781:2240030 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239781:2240030 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239781:2240030 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239782:2240031 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2239781:2240030 [0] NCCL INFO comm 0x102eb980 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2239782:2240031 [1] NCCL INFO comm 0x10b4d040 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239782:2240031 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2239782:2240031 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2239782:2240031 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2239782:2240031 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2239781:2240030 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2239781:2240030 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2239781:2240030 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2239781:2240030 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2239781:2240042 [0] NCCL INFO [Proxy Service] Device 0 CPU core 107 n136-128-154:2239781:2240043 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 108 n136-128-154:2239782:2240031 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2239782:2240044 [1] NCCL INFO [Proxy Service] Device 1 CPU core 40 n136-128-154:2239782:2240045 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 41 n136-128-154:2239781:2240030 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2239782:2240031 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2239781:2240030 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2239782:2240031 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2239781:2240030 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2239781:2240030 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2239781:2240030 [0] NCCL INFO ncclCommInitRankConfig comm 0x102eb980 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xc1a41f07abad06fa - Init COMPLETE n136-128-154:2239781:2240030 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.79 (kernels 0.12, alloc 0.43, bootstrap 0.00, allgathers 0.01, topo 0.07, graphs 0.00, connections 0.05, rest 0.12) n136-128-154:2239782:2240031 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2239782:2240031 [1] NCCL INFO ncclCommInitRankConfig comm 0x10b4d040 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xc1a41f07abad06fa - Init COMPLETE n136-128-154:2239782:2240031 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.78 (kernels 0.11, alloc 0.42, bootstrap 0.00, allgathers 0.01, topo 0.07, graphs 0.00, connections 0.05, rest 0.12) n136-128-154:2239781:2240046 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239781:2240046 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240047 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2239781:2240046 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:12:15 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:12:15 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/1268 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2239782:2240072 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2239782:2240076 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2239782:2240076 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2239782:2240076 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2239782:2240076 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2239782:2240076 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2239782:2240076 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2239782:2240044 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:12:46 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 5, batch_size: auto (64) | Tasks |Version|Filter|n-shot|Metric| |Value| |Stderr| |----------|------:|------|-----:|------|---|----:|---|-----:| |winogrande| 1|none | 5|acc |↑ |0.498|± |0.0141| [rank0]:[W1209 19:12:47.537418135 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2239781:2240127 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2239781:2240127 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2239781:2240127 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2239781:2240042 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2239781:2240127 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2239781:2240127 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2239781:2240127 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2239782:2240044 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2239782:2240076 [1] NCCL INFO comm 0x10b4d040 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2239781:2240127 [0] NCCL INFO comm 0x102eb980 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 winogrande 评估完成! ================================================== 开始评估:任务=boolq | 少样本数=0 | 模型=Qwen2.5-7B-quantization-fg 输出路径:results2/Qwen2.5-7B-quantization-fg/f6/boolq.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:19:13:38 INFO [__main__:440] Selected Tasks: ['boolq'] 2025-12-09:19:13:38 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:13:38 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:13:38 INFO [__main__:440] Selected Tasks: ['boolq'] 2025-12-09:19:13:38 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:13:38 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:13:39 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:13:39 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2240172:2240375 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2240172:2240375 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2240172:2240375 [0] NCCL INFO Using network IB n136-128-154:2240172:2240375 [0] NCCL INFO ncclCommInitRankConfig comm 0x11218200 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xe55dfc9887a40091 - Init START 2025-12-09:19:14:08 WARNING [evaluator:309] Overwriting default num_fewshot of boolq from None to 0 2025-12-09:19:14:08 INFO [api.task:434] Building contexts for boolq on rank 1... 0%| | 0/1635 [00:00 n136-128-154:2240173:2240390 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2240173:2240390 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2240173:2240390 [1] NCCL INFO Using network IB n136-128-154:2240173:2240390 [1] NCCL INFO ncclCommInitRankConfig comm 0xfab04c0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xe55dfc9887a40091 - Init START n136-128-154:2240173:2240390 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2240172:2240375 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2240173:2240390 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2240173:2240390 [1] NCCL INFO Retrieving state for IB n136-128-154:2240173:2240390 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2240173:2240390 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2240172:2240375 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2240172:2240375 [0] NCCL INFO Retrieving state for IB n136-128-154:2240172:2240375 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2240173:2240390 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2240172:2240375 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2240173:2240390 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2240172:2240375 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2240173:2240390 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2240172:2240375 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2240172:2240375 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2240173:2240390 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240173:2240390 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240172:2240375 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240172:2240375 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240172:2240375 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2240172:2240375 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2240172:2240375 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2240172:2240375 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2240172:2240375 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2240172:2240375 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2240172:2240375 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2240172:2240375 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2240172:2240375 [0] NCCL INFO ========================================== n136-128-154:2240172:2240375 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240172:2240375 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240172:2240375 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2240173:2240390 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2240173:2240390 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2240173:2240390 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2240173:2240390 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2240173:2240390 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2240173:2240390 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2240173:2240390 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2240173:2240390 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2240173:2240390 [1] NCCL INFO ========================================== n136-128-154:2240173:2240390 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240173:2240390 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240173:2240390 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2240172:2240375 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2240172:2240375 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2240172:2240375 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240172:2240375 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240172:2240375 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240172:2240375 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240172:2240375 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240172:2240375 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240172:2240375 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240173:2240390 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2240173:2240390 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2240173:2240390 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240173:2240390 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240173:2240390 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240173:2240390 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240173:2240390 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240173:2240390 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240173:2240390 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240172:2240375 [0] NCCL INFO comm 0x11218200 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2240173:2240390 [1] NCCL INFO comm 0xfab04c0 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240173:2240390 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2240173:2240390 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2240173:2240390 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2240173:2240390 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2240172:2240375 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2240172:2240375 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2240173:2240390 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2240173:2240397 [1] NCCL INFO [Proxy Service] Device 1 CPU core 119 n136-128-154:2240173:2240398 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 57 n136-128-154:2240172:2240375 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2240172:2240375 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2240172:2240399 [0] NCCL INFO [Proxy Service] Device 0 CPU core 107 n136-128-154:2240172:2240400 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 109 n136-128-154:2240173:2240390 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2240173:2240390 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2240172:2240375 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2240172:2240375 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2240172:2240375 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2240173:2240390 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2240173:2240390 [1] NCCL INFO ncclCommInitRankConfig comm 0xfab04c0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xe55dfc9887a40091 - Init COMPLETE n136-128-154:2240173:2240390 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.50 (kernels 0.11, alloc 0.20, bootstrap 0.00, allgathers 0.00, topo 0.06, graphs 0.00, connections 0.01, rest 0.12) n136-128-154:2240172:2240375 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2240172:2240375 [0] NCCL INFO ncclCommInitRankConfig comm 0x11218200 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xe55dfc9887a40091 - Init COMPLETE n136-128-154:2240172:2240375 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 2.12 (kernels 0.19, alloc 0.26, bootstrap 1.47, allgathers 0.00, topo 0.06, graphs 0.00, connections 0.01, rest 0.12) n136-128-154:2240173:2240401 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240172:2240402 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240173:2240401 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2240172:2240402 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:14:09 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:14:09 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/3270 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240173:2240489 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2240173:2240493 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240173:2240493 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240173:2240493 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240173:2240493 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240173:2240493 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240173:2240493 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240173:2240397 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:15:39 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: auto (45) |Tasks|Version|Filter|n-shot|Metric| |Value | |Stderr| |-----|------:|------|-----:|------|---|-----:|---|-----:| |boolq| 2|none | 0|acc |↑ |0.3801|± |0.0085| [rank0]:[W1209 19:15:39.023581183 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2240172:2240544 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240172:2240544 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240172:2240544 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240172:2240399 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2240172:2240544 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240172:2240544 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240172:2240544 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240173:2240397 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2240173:2240493 [1] NCCL INFO comm 0xfab04c0 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2240172:2240544 [0] NCCL INFO comm 0x11218200 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 boolq 评估完成! ================================================== 开始评估:任务=arc_challenge | 少样本数=25 | 模型=Qwen2.5-7B-quantization-fg 输出路径:results2/Qwen2.5-7B-quantization-fg/f6/arc_challenge.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:19:16:29 INFO [__main__:440] Selected Tasks: ['arc_challenge'] 2025-12-09:19:16:29 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:16:29 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:16:29 INFO [__main__:440] Selected Tasks: ['arc_challenge'] 2025-12-09:19:16:29 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:16:29 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:16:30 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:16:30 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:16:30 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'} `torch_dtype` is deprecated! Use `dtype` instead! `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2240589:2240786 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2240589:2240786 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2240589:2240786 [0] NCCL INFO Using network IB 93%|█████████▎| 544/586 [00:14<00:01, 36.65it/s]n136-128-154:2240589:2240786 [0] NCCL INFO ncclCommInitRankConfig comm 0x119f6810 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xc11a58f1b0081fd7 - Init START 94%|█████████▎| 548/586 [00:15<00:01, 36.76it/s] 94%|█████████▍| 552/586 [00:15<00:00, 36.80it/s] 95%|█████████▍| 556/586 [00:15<00:00, 36.79it/s] 96%|█████████▌| 560/586 [00:15<00:00, 36.72it/s] 96%|█████████▌| 564/586 [00:15<00:00, 36.67it/s] 97%|█████████▋| 568/586 [00:15<00:00, 36.58it/s] 98%|█████████▊| 572/586 [00:15<00:00, 36.58it/s] 98%|█████████▊| 576/586 [00:15<00:00, 36.56it/s] 99%|█████████▉| 580/586 [00:15<00:00, 36.58it/s] 100%|█████████▉| 584/586 [00:16<00:00, 36.61it/s] 100%|██████████| 586/586 [00:16<00:00, 36.44it/s] n136-128-154:2240590:2240590 [1] NCCL INFO cudaDriverVersion 12040 n136-128-154:2240590:2240590 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 n136-128-154:2240590:2240590 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 n136-128-154:2240590:2240590 [1] NCCL INFO NCCL version 2.27.5+cuda12.9 n136-128-154:2240590:2240590 [1] NCCL INFO Comm config Blocking set to 1 n136-128-154:2240590:2240791 [1] NCCL INFO NET/Plugin: Could not find: none libnccl-net-none.so. n136-128-154:2240590:2240791 [1] NCCL INFO NCCL_IB_DISABLE set by environment to 0. n136-128-154:2240590:2240791 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 n136-128-154:2240590:2240791 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 n136-128-154:2240590:2240791 [1] NCCL INFO NCCL_IB_HCA set to mlx5 n136-128-154:2240590:2240791 [1] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2240590:2240791 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2240590:2240791 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2240590:2240791 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2240590:2240791 [1] NCCL INFO Using network IB n136-128-154:2240590:2240791 [1] NCCL INFO ncclCommInitRankConfig comm 0x109605b0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xc11a58f1b0081fd7 - Init START n136-128-154:2240590:2240791 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2240589:2240786 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2240589:2240786 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2240589:2240786 [0] NCCL INFO Retrieving state for IB n136-128-154:2240589:2240786 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2240589:2240786 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2240589:2240786 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2240589:2240786 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2240589:2240786 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2240590:2240791 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2240590:2240791 [1] NCCL INFO Retrieving state for IB n136-128-154:2240590:2240791 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2240590:2240791 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2240590:2240791 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2240590:2240791 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2240590:2240791 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2240590:2240791 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240590:2240791 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240589:2240786 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240589:2240786 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2240589:2240786 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2240589:2240786 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2240589:2240786 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2240589:2240786 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2240589:2240786 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2240589:2240786 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2240589:2240786 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2240589:2240786 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2240589:2240786 [0] NCCL INFO ========================================== n136-128-154:2240589:2240786 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240589:2240786 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240589:2240786 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2240590:2240791 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2240590:2240791 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2240590:2240791 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2240590:2240791 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2240590:2240791 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2240590:2240791 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2240590:2240791 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2240590:2240791 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2240590:2240791 [1] NCCL INFO ========================================== n136-128-154:2240590:2240791 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240590:2240791 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2240590:2240791 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2240589:2240786 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2240589:2240786 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2240589:2240786 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240589:2240786 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240589:2240786 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240589:2240786 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240589:2240786 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240589:2240786 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240589:2240786 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2240590:2240791 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2240590:2240791 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2240590:2240791 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2240590:2240791 [1] NCCL INFO comm 0x109605b0 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO comm 0x119f6810 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2240589:2240786 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2240590:2240791 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2240590:2240791 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2240589:2240786 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2240589:2240786 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2240590:2240791 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2240590:2240800 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 55 n136-128-154:2240590:2240799 [1] NCCL INFO [Proxy Service] Device 1 CPU core 117 n136-128-154:2240589:2240786 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2240589:2240786 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2240589:2240802 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 56 n136-128-154:2240589:2240801 [0] NCCL INFO [Proxy Service] Device 0 CPU core 50 n136-128-154:2240590:2240791 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2240590:2240791 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2240589:2240786 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2240589:2240786 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2240589:2240786 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2240589:2240786 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2240589:2240786 [0] NCCL INFO ncclCommInitRankConfig comm 0x119f6810 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xc11a58f1b0081fd7 - Init COMPLETE n136-128-154:2240589:2240786 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 1.95 (kernels 0.11, alloc 0.17, bootstrap 1.50, allgathers 0.01, topo 0.06, graphs 0.00, connections 0.01, rest 0.10) n136-128-154:2240590:2240791 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2240590:2240791 [1] NCCL INFO ncclCommInitRankConfig comm 0x109605b0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xc11a58f1b0081fd7 - Init COMPLETE n136-128-154:2240590:2240791 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.52 (kernels 0.11, alloc 0.23, bootstrap 0.00, allgathers 0.01, topo 0.06, graphs 0.00, connections 0.01, rest 0.10) n136-128-154:2240589:2240803 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240589:2240803 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2240804 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2240589:2240803 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:17:07 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:17:07 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/2344 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2240590:2241157 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2240590:2241162 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240590:2241162 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240590:2241162 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240590:2241162 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240590:2241162 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240590:2241162 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240590:2240799 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:20:53 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 25, batch_size: auto (45) | Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr| |-------------|------:|------|-----:|--------|---|-----:|---|-----:| |arc_challenge| 1|none | 25|acc |↑ |0.2176|± |0.0121| | | |none | 25|acc_norm|↑ |0.2662|± |0.0129| [rank0]:[W1209 19:20:54.772226287 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2240589:2241211 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240589:2241211 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240589:2241211 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240589:2240801 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2240590:2240799 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2240589:2241211 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2240589:2241211 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2240589:2241211 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2240590:2241162 [1] NCCL INFO comm 0x109605b0 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2240589:2241211 [0] NCCL INFO comm 0x119f6810 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 arc_challenge 评估完成! ================================================== 开始评估:任务=truthfulqa_mc1 | 少样本数=0 | 模型=Qwen2.5-7B-quantization-fg 输出路径:results2/Qwen2.5-7B-quantization-fg/f6/truthfulqa_mc1.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:19:21:44 INFO [__main__:440] Selected Tasks: ['truthfulqa_mc1'] 2025-12-09:19:21:44 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:21:44 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:21:44 INFO [__main__:440] Selected Tasks: ['truthfulqa_mc1'] 2025-12-09:19:21:44 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:21:44 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:21:45 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:21:45 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2241273:2241459 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2241272:2241458 [0] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2241272:2241458 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2241272:2241458 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2241273:2241459 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2241273:2241459 [1] NCCL INFO Using network IB n136-128-154:2241272:2241458 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2241272:2241458 [0] NCCL INFO Using network IB n136-128-154:2241273:2241459 [1] NCCL INFO ncclCommInitRankConfig comm 0x10e35c90 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xcadda8d18b391fd7 - Init START n136-128-154:2241272:2241458 [0] NCCL INFO ncclCommInitRankConfig comm 0xf6198f0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xcadda8d18b391fd7 - Init START n136-128-154:2241273:2241459 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2241272:2241458 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2241273:2241459 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2241273:2241459 [1] NCCL INFO Retrieving state for IB n136-128-154:2241273:2241459 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2241273:2241459 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2241272:2241458 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2241272:2241458 [0] NCCL INFO Retrieving state for IB n136-128-154:2241272:2241458 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2241273:2241459 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2241272:2241458 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2241272:2241458 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2241273:2241459 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2241273:2241459 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2241272:2241458 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2241272:2241458 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2241272:2241458 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241272:2241458 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241272:2241458 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2241272:2241458 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2241272:2241458 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2241272:2241458 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2241272:2241458 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2241272:2241458 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2241272:2241458 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2241272:2241458 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2241272:2241458 [0] NCCL INFO ========================================== n136-128-154:2241272:2241458 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241272:2241458 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241272:2241458 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2241272:2241458 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2241272:2241458 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2241272:2241458 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241273:2241459 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241273:2241459 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2241273:2241459 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2241273:2241459 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2241273:2241459 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2241273:2241459 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2241273:2241459 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2241273:2241459 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2241273:2241459 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2241273:2241459 [1] NCCL INFO ========================================== n136-128-154:2241273:2241459 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241273:2241459 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241273:2241459 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2241273:2241459 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2241273:2241459 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2241273:2241459 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241273:2241459 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241273:2241459 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241273:2241459 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241273:2241459 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241273:2241459 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241273:2241459 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241272:2241458 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241272:2241458 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241272:2241458 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241272:2241458 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241272:2241458 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241272:2241458 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241272:2241458 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241273:2241459 [1] NCCL INFO comm 0x10e35c90 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2241272:2241458 [0] NCCL INFO comm 0xf6198f0 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241272:2241458 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2241273:2241459 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2241272:2241458 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2241272:2241458 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2241272:2241458 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2241273:2241459 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2241273:2241471 [1] NCCL INFO [Proxy Service] Device 1 CPU core 61 n136-128-154:2241273:2241472 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 37 n136-128-154:2241272:2241458 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2241272:2241458 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2241272:2241473 [0] NCCL INFO [Proxy Service] Device 0 CPU core 99 n136-128-154:2241272:2241474 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 103 n136-128-154:2241273:2241459 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2241273:2241459 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2241272:2241458 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2241272:2241458 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2241272:2241458 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2241272:2241458 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2241272:2241458 [0] NCCL INFO ncclCommInitRankConfig comm 0xf6198f0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xcadda8d18b391fd7 - Init COMPLETE n136-128-154:2241272:2241458 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.77 (kernels 0.12, alloc 0.42, bootstrap 0.00, allgathers 0.00, topo 0.06, graphs 0.00, connections 0.03, rest 0.12) n136-128-154:2241273:2241459 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2241273:2241459 [1] NCCL INFO ncclCommInitRankConfig comm 0x10e35c90 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xcadda8d18b391fd7 - Init COMPLETE n136-128-154:2241273:2241459 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.76 (kernels 0.12, alloc 0.42, bootstrap 0.00, allgathers 0.00, topo 0.06, graphs 0.00, connections 0.02, rest 0.14) n136-128-154:2241272:2241475 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241476 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241272:2241475 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2241273:2241476 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:22:12 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:22:12 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/2066 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241273:2241528 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2241273:2241532 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241273:2241532 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241273:2241532 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241273:2241532 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241273:2241532 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241273:2241532 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241273:2241471 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:23:03 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: auto (64) | Tasks |Version|Filter|n-shot|Metric| |Value | |Stderr| |--------------|------:|------|-----:|------|---|-----:|---|-----:| |truthfulqa_mc1| 2|none | 0|acc |↑ |0.2387|± |0.0149| [rank0]:[W1209 19:23:04.417594440 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2241272:2241583 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241272:2241583 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241272:2241583 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241272:2241473 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2241272:2241583 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241272:2241583 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241272:2241583 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241273:2241471 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2241273:2241532 [1] NCCL INFO comm 0x10e35c90 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2241272:2241583 [0] NCCL INFO comm 0xf6198f0 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 truthfulqa_mc1 评估完成! ================================================== 开始评估:任务=piqa | 少样本数=0 | 模型=Qwen2.5-7B-quantization-fg 输出路径:results2/Qwen2.5-7B-quantization-fg/f6/piqa.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:19:23:54 INFO [__main__:440] Selected Tasks: ['piqa'] 2025-12-09:19:23:54 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:23:54 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:23:54 INFO [__main__:440] Selected Tasks: ['piqa'] 2025-12-09:19:23:54 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:23:54 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:23:55 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:23:55 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'} `torch_dtype` is deprecated! Use `dtype` instead! The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:23:55 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2241630:2241849 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2241630:2241849 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2241630:2241849 [0] NCCL INFO Using network IB n136-128-154:2241630:2241849 [0] NCCL INFO ncclCommInitRankConfig comm 0xe995780 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x53eacb4832208b8c - Init START 2025-12-09:19:24:16 WARNING [evaluator:309] Overwriting default num_fewshot of piqa from None to 0 2025-12-09:19:24:16 INFO [api.task:434] Building contexts for piqa on rank 1... 0%| | 0/919 [00:00 n136-128-154:2241631:2241886 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2241631:2241886 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2241631:2241886 [1] NCCL INFO Using network IB n136-128-154:2241631:2241886 [1] NCCL INFO ncclCommInitRankConfig comm 0xea03000 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x53eacb4832208b8c - Init START n136-128-154:2241631:2241886 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2241630:2241849 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2241630:2241849 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2241630:2241849 [0] NCCL INFO Retrieving state for IB n136-128-154:2241630:2241849 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2241630:2241849 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2241631:2241886 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2241631:2241886 [1] NCCL INFO Retrieving state for IB n136-128-154:2241631:2241886 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2241630:2241849 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2241631:2241886 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2241630:2241849 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2241631:2241886 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2241630:2241849 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2241631:2241886 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2241631:2241886 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2241630:2241849 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241630:2241849 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241631:2241886 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241631:2241886 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2241630:2241849 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2241630:2241849 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2241630:2241849 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2241630:2241849 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2241630:2241849 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2241630:2241849 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2241630:2241849 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2241630:2241849 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2241630:2241849 [0] NCCL INFO ========================================== n136-128-154:2241630:2241849 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241630:2241849 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241630:2241849 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2241631:2241886 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2241631:2241886 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2241631:2241886 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2241631:2241886 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2241631:2241886 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2241631:2241886 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2241631:2241886 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2241631:2241886 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2241631:2241886 [1] NCCL INFO ========================================== n136-128-154:2241631:2241886 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241631:2241886 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2241631:2241886 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2241630:2241849 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2241630:2241849 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2241630:2241849 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241630:2241849 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241630:2241849 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241630:2241849 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241630:2241849 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241630:2241849 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241630:2241849 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241631:2241886 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2241631:2241886 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2241631:2241886 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2241631:2241886 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241631:2241886 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241631:2241886 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241631:2241886 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241631:2241886 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241631:2241886 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2241630:2241849 [0] NCCL INFO comm 0xe995780 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2241631:2241886 [1] NCCL INFO comm 0xea03000 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241631:2241886 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2241631:2241886 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2241631:2241886 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2241631:2241886 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2241630:2241849 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2241630:2241849 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2241631:2241886 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2241631:2241895 [1] NCCL INFO [Proxy Service] Device 1 CPU core 54 n136-128-154:2241631:2241896 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 55 n136-128-154:2241630:2241849 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2241630:2241849 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2241630:2241897 [0] NCCL INFO [Proxy Service] Device 0 CPU core 35 n136-128-154:2241630:2241898 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 37 n136-128-154:2241631:2241886 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2241631:2241886 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2241630:2241849 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2241630:2241849 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2241630:2241849 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2241631:2241886 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2241631:2241886 [1] NCCL INFO ncclCommInitRankConfig comm 0xea03000 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x53eacb4832208b8c - Init COMPLETE n136-128-154:2241631:2241886 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.70 (kernels 0.37, alloc 0.19, bootstrap 0.00, allgathers 0.00, topo 0.10, graphs 0.00, connections 0.01, rest 0.03) n136-128-154:2241630:2241849 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2241630:2241849 [0] NCCL INFO ncclCommInitRankConfig comm 0xe995780 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x53eacb4832208b8c - Init COMPLETE n136-128-154:2241630:2241849 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 3.25 (kernels 0.20, alloc 0.11, bootstrap 2.80, allgathers 0.00, topo 0.10, graphs 0.00, connections 0.01, rest 0.03) n136-128-154:2241631:2241899 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241630:2241900 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2241631:2241899 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2241630:2241900 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:24:18 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:24:18 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/1838 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2241631:2242079 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2241631:2242083 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241631:2242083 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241631:2242083 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241631:2242083 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241631:2242083 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241631:2242083 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241631:2241895 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:24:46 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: auto (64) |Tasks|Version|Filter|n-shot| Metric | |Value | |Stderr| |-----|------:|------|-----:|--------|---|-----:|---|-----:| |piqa | 1|none | 0|acc |↑ |0.5332|± |0.0116| | | |none | 0|acc_norm|↑ |0.5152|± |0.0117| [rank0]:[W1209 19:24:47.098294767 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2241630:2242193 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241630:2242193 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241630:2242193 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241630:2241897 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2241631:2241895 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2241630:2242193 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2241630:2242193 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2241630:2242193 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2241631:2242083 [1] NCCL INFO comm 0xea03000 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2241630:2242193 [0] NCCL INFO comm 0xe995780 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 piqa 评估完成! ================================================== 开始评估:任务=hellaswag | 少样本数=10 | 模型=Qwen2.5-7B-quantization-fg 输出路径:results2/Qwen2.5-7B-quantization-fg/f6/hellaswag.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:19:25:39 INFO [__main__:440] Selected Tasks: ['hellaswag'] 2025-12-09:19:25:39 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:25:39 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:25:39 INFO [__main__:440] Selected Tasks: ['hellaswag'] 2025-12-09:19:25:39 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:25:39 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg'} 2025-12-09:19:25:40 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:25:40 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2242239:2242479 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2242239:2242479 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2242239:2242479 [0] NCCL INFO Using network IB n136-128-154:2242240:2242480 [1] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2242240:2242480 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2242240:2242480 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2242240:2242480 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2242240:2242480 [1] NCCL INFO Using network IB n136-128-154:2242240:2242480 [1] NCCL INFO ncclCommInitRankConfig comm 0x47be4560 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x624137be80e2f945 - Init START n136-128-154:2242239:2242479 [0] NCCL INFO ncclCommInitRankConfig comm 0xe2505d0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x624137be80e2f945 - Init START n136-128-154:2242239:2242479 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2242240:2242480 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2242240:2242480 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2242240:2242480 [1] NCCL INFO Retrieving state for IB n136-128-154:2242240:2242480 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2242240:2242480 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2242239:2242479 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2242239:2242479 [0] NCCL INFO Retrieving state for IB n136-128-154:2242239:2242479 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2242240:2242480 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2242239:2242479 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2242240:2242480 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2242239:2242479 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2242240:2242480 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2242239:2242479 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2242239:2242479 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2242240:2242480 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2242240:2242480 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2242239:2242479 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2242239:2242479 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2242240:2242480 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2242240:2242480 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2242240:2242480 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2242240:2242480 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2242240:2242480 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2242240:2242480 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2242240:2242480 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2242240:2242480 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2242240:2242480 [1] NCCL INFO ========================================== n136-128-154:2242240:2242480 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2242240:2242480 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2242239:2242479 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2242239:2242479 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2242240:2242480 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2242239:2242479 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2242239:2242479 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2242239:2242479 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2242239:2242479 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2242239:2242479 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2242239:2242479 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2242239:2242479 [0] NCCL INFO ========================================== n136-128-154:2242239:2242479 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2242239:2242479 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2242239:2242479 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2242239:2242479 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2242239:2242479 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2242239:2242479 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2242240:2242480 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2242240:2242480 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242239:2242479 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2242240:2242480 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242239:2242479 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2242240:2242480 [1] NCCL INFO comm 0x47be4560 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2242239:2242479 [0] NCCL INFO comm 0xe2505d0 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242240:2242480 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2242240:2242480 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2242240:2242480 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2242240:2242480 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2242239:2242479 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2242239:2242479 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2242239:2242479 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2242239:2242479 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2242239:2242492 [0] NCCL INFO [Proxy Service] Device 0 CPU core 99 n136-128-154:2242239:2242493 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 100 n136-128-154:2242240:2242480 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2242240:2242494 [1] NCCL INFO [Proxy Service] Device 1 CPU core 104 n136-128-154:2242240:2242495 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 39 n136-128-154:2242240:2242480 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2242240:2242480 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2242239:2242479 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2242239:2242479 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2242239:2242479 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2242240:2242480 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2242240:2242480 [1] NCCL INFO ncclCommInitRankConfig comm 0x47be4560 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x624137be80e2f945 - Init COMPLETE n136-128-154:2242240:2242480 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.86 (kernels 0.14, alloc 0.47, bootstrap 0.00, allgathers 0.00, topo 0.07, graphs 0.00, connections 0.02, rest 0.16) n136-128-154:2242239:2242479 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2242239:2242479 [0] NCCL INFO ncclCommInitRankConfig comm 0xe2505d0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x624137be80e2f945 - Init COMPLETE n136-128-154:2242239:2242479 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.87 (kernels 0.13, alloc 0.48, bootstrap 0.00, allgathers 0.00, topo 0.07, graphs 0.00, connections 0.02, rest 0.16) n136-128-154:2242240:2242496 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242239:2242497 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2242240:2242496 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2242239:2242497 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:26:34 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:26:34 INFO [evaluator:559] Running loglikelihood requests Passed argument batch_size = auto:1. Detecting largest batch size Running loglikelihood requests: 0%| | 0/20084 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2242240:2245500 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2242240:2245537 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2242240:2245537 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2242240:2245537 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2242240:2245537 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2242240:2245537 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2242240:2245537 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2242240:2242494 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:50:17 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen2.5-7B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 10, batch_size: auto (57) | Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr| |---------|------:|------|-----:|--------|---|-----:|---|-----:| |hellaswag| 1|none | 10|acc |↑ |0.2590|± |0.0044| | | |none | 10|acc_norm|↑ |0.2554|± |0.0044| [rank0]:[W1209 19:50:18.642585790 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2242239:2245602 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2242239:2245602 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2242239:2245602 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2242239:2242492 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2242240:2242494 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2242239:2245602 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2242239:2245602 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2242239:2245602 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2242240:2245537 [1] NCCL INFO comm 0x47be4560 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2242239:2245602 [0] NCCL INFO comm 0xe2505d0 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 hellaswag 评估完成! Execution time: 2673.92 seconds Namespace(save_dir='/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg', self_attn_layer_to_quant='29 23 24 30 18 28 26 20 16 27 25 17 19 21', mlp_layer_to_quant='26 20 22', model_id='/mnt/bn/life-mllm/users/cxr/quantization/models/meta-llama/Llama-3.1-8B', cuda_id=6) `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2246512:2246882 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2246513:2246883 [1] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2246513:2246883 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2246513:2246883 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2246512:2246882 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2246512:2246882 [0] NCCL INFO Using network IB n136-128-154:2246513:2246883 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2246513:2246883 [1] NCCL INFO Using network IB n136-128-154:2246512:2246882 [0] NCCL INFO ncclCommInitRankConfig comm 0x14c90df0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xa8d5bc1f63708c6 - Init START n136-128-154:2246513:2246883 [1] NCCL INFO ncclCommInitRankConfig comm 0x16758670 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xa8d5bc1f63708c6 - Init START n136-128-154:2246512:2246882 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2246513:2246883 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2246512:2246882 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2246512:2246882 [0] NCCL INFO Retrieving state for IB n136-128-154:2246512:2246882 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2246512:2246882 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2246513:2246883 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2246513:2246883 [1] NCCL INFO Retrieving state for IB n136-128-154:2246513:2246883 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2246512:2246882 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2246513:2246883 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2246512:2246882 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2246513:2246883 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2246512:2246882 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2246513:2246883 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2246513:2246883 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2246513:2246883 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2246513:2246883 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2246512:2246882 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2246512:2246882 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2246513:2246883 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2246513:2246883 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2246513:2246883 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2246513:2246883 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2246513:2246883 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2246513:2246883 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2246513:2246883 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2246512:2246882 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2246513:2246883 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2246512:2246882 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2246513:2246883 [1] NCCL INFO ========================================== n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2246513:2246883 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2246513:2246883 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2246512:2246882 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2246512:2246882 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2246513:2246883 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2246512:2246882 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2246512:2246882 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2246512:2246882 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2246512:2246882 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2246512:2246882 [0] NCCL INFO ========================================== n136-128-154:2246512:2246882 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2246512:2246882 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2246512:2246882 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2246513:2246883 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2246513:2246883 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2246513:2246883 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246513:2246883 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246513:2246883 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246513:2246883 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246513:2246883 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246513:2246883 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246513:2246883 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246512:2246882 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2246512:2246882 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2246512:2246882 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2246512:2246882 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246512:2246882 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246512:2246882 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246512:2246882 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246512:2246882 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246512:2246882 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2246513:2246883 [1] NCCL INFO comm 0x16758670 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2246512:2246882 [0] NCCL INFO comm 0x14c90df0 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246513:2246883 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2246513:2246883 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2246513:2246883 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2246513:2246883 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2246512:2246882 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2246512:2246882 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2246513:2246883 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2246513:2246894 [1] NCCL INFO [Proxy Service] Device 1 CPU core 109 n136-128-154:2246513:2246895 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 110 n136-128-154:2246512:2246882 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2246512:2246882 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2246512:2246896 [0] NCCL INFO [Proxy Service] Device 0 CPU core 99 n136-128-154:2246512:2246897 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 36 n136-128-154:2246513:2246883 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2246513:2246883 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2246512:2246882 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2246512:2246882 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2246512:2246882 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2246513:2246883 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2246513:2246883 [1] NCCL INFO ncclCommInitRankConfig comm 0x16758670 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xa8d5bc1f63708c6 - Init COMPLETE n136-128-154:2246513:2246883 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.41 (kernels 0.12, alloc 0.18, bootstrap 0.00, allgathers 0.00, topo 0.07, graphs 0.00, connections 0.01, rest 0.03) n136-128-154:2246512:2246882 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2246512:2246882 [0] NCCL INFO ncclCommInitRankConfig comm 0x14c90df0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xa8d5bc1f63708c6 - Init COMPLETE n136-128-154:2246512:2246882 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.41 (kernels 0.13, alloc 0.18, bootstrap 0.00, allgathers 0.00, topo 0.07, graphs 0.00, connections 0.01, rest 0.03) n136-128-154:2246513:2246898 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246512:2246899 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2246513:2246898 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2246512:2246899 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:56:28 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:56:28 INFO [evaluator:559] Running loglikelihood requests Passed argument batch_size = auto:1. Detecting largest batch size Running loglikelihood requests: 0%| | 0/1268 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2246513:2247195 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2246513:2247200 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2246513:2247200 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2246513:2247200 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2246513:2247200 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2246513:2247200 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2246513:2247200 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2246513:2246894 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:57:00 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 5, batch_size: auto (64) | Tasks |Version|Filter|n-shot|Metric| |Value | |Stderr| |----------|------:|------|-----:|------|---|-----:|---|-----:| |winogrande| 1|none | 5|acc |↑ |0.6693|± |0.0132| [rank0]:[W1209 19:57:00.913929289 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2246512:2247311 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2246512:2247311 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2246512:2247311 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2246512:2246896 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2246512:2247311 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2246512:2247311 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2246512:2247311 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2246513:2246894 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2246513:2247200 [1] NCCL INFO comm 0x16758670 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2246512:2247311 [0] NCCL INFO comm 0x14c90df0 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 winogrande 评估完成! ================================================== 开始评估:任务=boolq | 少样本数=0 | 模型=Llama-3.1-8B-quantization-fg 输出路径:results2/Llama-3.1-8B-quantization-fg/fg6/boolq.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:19:57:52 INFO [__main__:440] Selected Tasks: ['boolq'] 2025-12-09:19:57:52 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:57:52 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:19:57:52 INFO [__main__:440] Selected Tasks: ['boolq'] 2025-12-09:19:57:52 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:19:57:52 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:19:57:53 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:19:57:53 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2247419:2247676 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2247419:2247676 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2247419:2247676 [0] NCCL INFO Using network IB n136-128-154:2247420:2247677 [1] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2247420:2247677 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2247420:2247677 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2247420:2247677 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2247420:2247677 [1] NCCL INFO Using network IB n136-128-154:2247419:2247676 [0] NCCL INFO ncclCommInitRankConfig comm 0x13269030 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xc5295f96de61b1e1 - Init START n136-128-154:2247420:2247677 [1] NCCL INFO ncclCommInitRankConfig comm 0xfc4f9e0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xc5295f96de61b1e1 - Init START n136-128-154:2247419:2247676 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2247420:2247677 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2247420:2247677 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2247420:2247677 [1] NCCL INFO Retrieving state for IB n136-128-154:2247420:2247677 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2247420:2247677 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2247419:2247676 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2247419:2247676 [0] NCCL INFO Retrieving state for IB n136-128-154:2247419:2247676 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2247420:2247677 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2247419:2247676 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2247420:2247677 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2247419:2247676 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2247420:2247677 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2247419:2247676 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2247419:2247676 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2247420:2247677 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247420:2247677 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247419:2247676 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247419:2247676 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247420:2247677 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2247420:2247677 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2247419:2247676 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2247419:2247676 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2247420:2247677 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2247419:2247676 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247420:2247677 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2247420:2247677 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2247419:2247676 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247420:2247677 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2247419:2247676 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2247419:2247676 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2247420:2247677 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2247420:2247677 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2247420:2247677 [1] NCCL INFO ========================================== n136-128-154:2247419:2247676 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2247419:2247676 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2247419:2247676 [0] NCCL INFO ========================================== n136-128-154:2247420:2247677 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247419:2247676 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247420:2247677 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247419:2247676 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247420:2247677 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2247419:2247676 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2247420:2247677 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2247420:2247677 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2247420:2247677 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247420:2247677 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247420:2247677 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247420:2247677 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247420:2247677 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247420:2247677 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247420:2247677 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247419:2247676 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2247419:2247676 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2247419:2247676 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247419:2247676 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247419:2247676 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247419:2247676 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247419:2247676 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247419:2247676 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247419:2247676 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247420:2247677 [1] NCCL INFO comm 0xfc4f9e0 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2247419:2247676 [0] NCCL INFO comm 0x13269030 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247420:2247677 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2247420:2247677 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2247420:2247677 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2247420:2247677 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2247419:2247676 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2247419:2247676 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2247420:2247677 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2247420:2247689 [1] NCCL INFO [Proxy Service] Device 1 CPU core 102 n136-128-154:2247420:2247690 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 39 n136-128-154:2247419:2247676 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2247419:2247676 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2247419:2247691 [0] NCCL INFO [Proxy Service] Device 0 CPU core 122 n136-128-154:2247419:2247692 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 123 n136-128-154:2247420:2247677 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2247420:2247677 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2247419:2247676 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2247419:2247676 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2247419:2247676 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2247420:2247677 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2247420:2247677 [1] NCCL INFO ncclCommInitRankConfig comm 0xfc4f9e0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xc5295f96de61b1e1 - Init COMPLETE n136-128-154:2247420:2247677 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.40 (kernels 0.13, alloc 0.17, bootstrap 0.00, allgathers 0.00, topo 0.05, graphs 0.00, connections 0.01, rest 0.04) n136-128-154:2247419:2247676 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2247419:2247676 [0] NCCL INFO ncclCommInitRankConfig comm 0x13269030 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xc5295f96de61b1e1 - Init COMPLETE n136-128-154:2247419:2247676 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.40 (kernels 0.13, alloc 0.17, bootstrap 0.00, allgathers 0.00, topo 0.05, graphs 0.00, connections 0.01, rest 0.04) n136-128-154:2247419:2247693 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247694 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247419:2247693 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2247420:2247694 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:19:58:21 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:19:58:21 INFO [evaluator:559] Running loglikelihood requests Passed argument batch_size = auto:1. Detecting largest batch size Running loglikelihood requests: 0%| | 0/3270 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247420:2247798 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2247420:2247804 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247420:2247804 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247420:2247804 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247420:2247804 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247420:2247804 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247420:2247804 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247420:2247689 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:19:59:31 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: auto (64) |Tasks|Version|Filter|n-shot|Metric| |Value | |Stderr| |-----|------:|------|-----:|------|---|-----:|---|-----:| |boolq| 2|none | 0|acc |↑ |0.7202|± |0.0079| [rank0]:[W1209 19:59:31.887913696 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2247419:2247891 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247419:2247891 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247419:2247891 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247419:2247691 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2247419:2247891 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247419:2247891 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247419:2247891 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247420:2247689 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2247420:2247804 [1] NCCL INFO comm 0xfc4f9e0 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2247419:2247891 [0] NCCL INFO comm 0x13269030 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 boolq 评估完成! ================================================== 开始评估:任务=arc_challenge | 少样本数=25 | 模型=Llama-3.1-8B-quantization-fg 输出路径:results2/Llama-3.1-8B-quantization-fg/fg6/arc_challenge.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:20:00:34 INFO [__main__:440] Selected Tasks: ['arc_challenge'] 2025-12-09:20:00:34 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:00:34 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:00:34 INFO [__main__:440] Selected Tasks: ['arc_challenge'] 2025-12-09:20:00:34 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:00:34 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:00:37 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:20:00:37 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2247972:2248484 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2247972:2248484 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2247972:2248484 [1] NCCL INFO Using network IB n136-128-154:2247970:2248483 [0] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2247970:2248483 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2247970:2248483 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2247970:2248483 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2247970:2248483 [0] NCCL INFO Using network IB n136-128-154:2247970:2248483 [0] NCCL INFO ncclCommInitRankConfig comm 0x127dd0e0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x306930abc4c7a324 - Init START n136-128-154:2247972:2248484 [1] NCCL INFO ncclCommInitRankConfig comm 0x112e94a0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x306930abc4c7a324 - Init START n136-128-154:2247970:2248483 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2247972:2248484 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2247970:2248483 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2247970:2248483 [0] NCCL INFO Retrieving state for IB n136-128-154:2247970:2248483 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2247970:2248483 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2247972:2248484 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2247972:2248484 [1] NCCL INFO Retrieving state for IB n136-128-154:2247972:2248484 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2247970:2248483 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2247972:2248484 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2247970:2248483 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2247972:2248484 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2247970:2248483 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2247972:2248484 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2247972:2248484 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2247970:2248483 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247970:2248483 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247972:2248484 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247972:2248484 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2247972:2248484 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2247972:2248484 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2247972:2248484 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2247972:2248484 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2247972:2248484 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2247972:2248484 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2247972:2248484 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2247972:2248484 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2247972:2248484 [1] NCCL INFO ========================================== n136-128-154:2247972:2248484 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247972:2248484 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247972:2248484 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2247970:2248483 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2247970:2248483 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2247970:2248483 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2247970:2248483 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2247970:2248483 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2247970:2248483 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2247970:2248483 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2247970:2248483 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2247970:2248483 [0] NCCL INFO ========================================== n136-128-154:2247970:2248483 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247970:2248483 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2247970:2248483 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2247972:2248484 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2247972:2248484 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2247970:2248483 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2247972:2248484 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247972:2248484 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2247970:2248483 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2247970:2248483 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2247970:2248483 [0] NCCL INFO comm 0x127dd0e0 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2247972:2248484 [1] NCCL INFO comm 0x112e94a0 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2247972:2248484 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2247972:2248484 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2247972:2248484 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2247972:2248484 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2247970:2248483 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2247970:2248483 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2247970:2248483 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2247970:2248483 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2247970:2248498 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 115 n136-128-154:2247970:2248497 [0] NCCL INFO [Proxy Service] Device 0 CPU core 113 n136-128-154:2247972:2248484 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2247972:2248499 [1] NCCL INFO [Proxy Service] Device 1 CPU core 52 n136-128-154:2247972:2248500 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 53 n136-128-154:2247970:2248483 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2247970:2248483 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2247972:2248484 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2247972:2248484 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2247970:2248483 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2247970:2248483 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2247970:2248483 [0] NCCL INFO ncclCommInitRankConfig comm 0x127dd0e0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x306930abc4c7a324 - Init COMPLETE n136-128-154:2247970:2248483 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.63 (kernels 0.12, alloc 0.30, bootstrap 0.00, allgathers 0.00, topo 0.05, graphs 0.00, connections 0.07, rest 0.08) n136-128-154:2247972:2248484 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2247972:2248484 [1] NCCL INFO ncclCommInitRankConfig comm 0x112e94a0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x306930abc4c7a324 - Init COMPLETE n136-128-154:2247972:2248484 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.62 (kernels 0.12, alloc 0.29, bootstrap 0.00, allgathers 0.00, topo 0.05, graphs 0.00, connections 0.07, rest 0.08) n136-128-154:2247970:2248501 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247970:2248501 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2248502 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2247970:2248501 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:20:01:17 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:20:01:17 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/2344 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2247972:2249091 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2247972:2249096 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247972:2249096 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247972:2249096 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247972:2249096 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247972:2249096 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247972:2249096 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247972:2248499 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:20:04:57 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 25, batch_size: auto (64) | Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr| |-------------|------:|------|-----:|--------|---|-----:|---|-----:| |arc_challenge| 1|none | 25|acc |↑ |0.3089|± |0.0135| | | |none | 25|acc_norm|↑ |0.3439|± |0.0139| [rank0]:[W1209 20:04:58.604574678 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2247970:2249190 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247970:2249190 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247970:2249190 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247970:2248497 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2247970:2249190 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2247970:2249190 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2247970:2249190 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2247972:2248499 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2247972:2249096 [1] NCCL INFO comm 0x112e94a0 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2247970:2249190 [0] NCCL INFO comm 0x127dd0e0 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 arc_challenge 评估完成! ================================================== 开始评估:任务=truthfulqa_mc1 | 少样本数=0 | 模型=Llama-3.1-8B-quantization-fg 输出路径:results2/Llama-3.1-8B-quantization-fg/fg6/truthfulqa_mc1.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:20:05:48 INFO [__main__:440] Selected Tasks: ['truthfulqa_mc1'] 2025-12-09:20:05:48 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:05:48 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:05:48 INFO [__main__:440] Selected Tasks: ['truthfulqa_mc1'] 2025-12-09:20:05:48 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:05:48 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:05:49 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:20:05:49 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2249344:2249774 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2249344:2249774 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2249344:2249774 [0] NCCL INFO Using network IB n136-128-154:2249344:2249774 [0] NCCL INFO ncclCommInitRankConfig comm 0x12ab5e20 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xdc5d9a9594c2fed0 - Init START 2025-12-09:20:06:17 INFO [evaluator:305] num_fewshot has been set to 0 for truthfulqa_mc1 in its config. Manual configuration will be ignored. 2025-12-09:20:06:17 INFO [api.task:434] Building contexts for truthfulqa_mc1 on rank 1... 0%| | 0/408 [00:00 n136-128-154:2249345:2249848 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2249345:2249848 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2249345:2249848 [1] NCCL INFO Using network IB n136-128-154:2249345:2249848 [1] NCCL INFO ncclCommInitRankConfig comm 0x151d1bb0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xdc5d9a9594c2fed0 - Init START n136-128-154:2249345:2249848 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2249344:2249774 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2249344:2249774 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2249344:2249774 [0] NCCL INFO Retrieving state for IB n136-128-154:2249344:2249774 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2249344:2249774 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2249345:2249848 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2249345:2249848 [1] NCCL INFO Retrieving state for IB n136-128-154:2249345:2249848 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2249344:2249774 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2249345:2249848 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2249344:2249774 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2249345:2249848 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2249344:2249774 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2249345:2249848 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2249345:2249848 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2249345:2249848 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2249345:2249848 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2249344:2249774 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2249344:2249774 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2249345:2249848 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2249345:2249848 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2249345:2249848 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2249345:2249848 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2249345:2249848 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2249345:2249848 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2249345:2249848 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2249345:2249848 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2249345:2249848 [1] NCCL INFO ========================================== n136-128-154:2249345:2249848 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2249345:2249848 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2249345:2249848 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2249345:2249848 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2249345:2249848 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2249345:2249848 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249345:2249848 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249345:2249848 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249345:2249848 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249345:2249848 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249345:2249848 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249345:2249848 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249344:2249774 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2249344:2249774 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2249344:2249774 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2249344:2249774 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2249344:2249774 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2249344:2249774 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2249344:2249774 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2249344:2249774 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2249344:2249774 [0] NCCL INFO ========================================== n136-128-154:2249344:2249774 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2249344:2249774 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2249344:2249774 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2249344:2249774 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2249344:2249774 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2249344:2249774 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2249344:2249774 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249344:2249774 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249344:2249774 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249344:2249774 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249344:2249774 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249344:2249774 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2249345:2249848 [1] NCCL INFO comm 0x151d1bb0 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2249344:2249774 [0] NCCL INFO comm 0x12ab5e20 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249345:2249848 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2249344:2249774 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2249345:2249848 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2249345:2249848 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2249345:2249848 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2249344:2249774 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2249344:2249774 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2249345:2249848 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2249345:2249857 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 103 n136-128-154:2249345:2249856 [1] NCCL INFO [Proxy Service] Device 1 CPU core 101 n136-128-154:2249344:2249774 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2249344:2249774 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2249344:2249858 [0] NCCL INFO [Proxy Service] Device 0 CPU core 104 n136-128-154:2249344:2249859 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 41 n136-128-154:2249345:2249848 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2249345:2249848 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2249344:2249774 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2249344:2249774 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2249344:2249774 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2249345:2249848 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2249345:2249848 [1] NCCL INFO ncclCommInitRankConfig comm 0x151d1bb0 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0xdc5d9a9594c2fed0 - Init COMPLETE n136-128-154:2249345:2249848 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.47 (kernels 0.13, alloc 0.18, bootstrap 0.00, allgathers 0.01, topo 0.09, graphs 0.00, connections 0.02, rest 0.04) n136-128-154:2249344:2249774 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2249344:2249774 [0] NCCL INFO ncclCommInitRankConfig comm 0x12ab5e20 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0xdc5d9a9594c2fed0 - Init COMPLETE n136-128-154:2249344:2249774 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 6.41 (kernels 0.12, alloc 0.17, bootstrap 5.96, allgathers 0.01, topo 0.09, graphs 0.00, connections 0.02, rest 0.05) n136-128-154:2249345:2249860 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249344:2249861 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2249345:2249860 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2249344:2249861 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:20:06:18 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:20:06:18 INFO [evaluator:559] Running loglikelihood requests Running loglikelihood requests: 0%| | 0/2066 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2249345:2249988 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2249345:2249994 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2249345:2249994 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2249345:2249994 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2249345:2249994 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2249345:2249994 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2249345:2249994 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2249345:2249856 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:20:07:09 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: auto (64) | Tasks |Version|Filter|n-shot|Metric| |Value | |Stderr| |--------------|------:|------|-----:|------|---|-----:|---|-----:| |truthfulqa_mc1| 2|none | 0|acc |↑ |0.2448|± |0.0151| [rank0]:[W1209 20:07:10.511964655 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2249344:2250056 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2249344:2250056 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2249344:2250056 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2249344:2249858 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2249344:2250056 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2249344:2250056 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2249344:2250056 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2249345:2249856 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2249345:2249994 [1] NCCL INFO comm 0x151d1bb0 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2249344:2250056 [0] NCCL INFO comm 0x12ab5e20 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 truthfulqa_mc1 评估完成! ================================================== 开始评估:任务=piqa | 少样本数=0 | 模型=Llama-3.1-8B-quantization-fg 输出路径:results2/Llama-3.1-8B-quantization-fg/fg6/piqa.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:20:08:00 INFO [__main__:440] Selected Tasks: ['piqa'] 2025-12-09:20:08:00 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:08:00 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:08:00 INFO [__main__:440] Selected Tasks: ['piqa'] 2025-12-09:20:08:00 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:08:00 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:08:01 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:20:08:01 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:1'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2250248:2250743 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2250248:2250743 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2250248:2250743 [1] NCCL INFO Using network IB n136-128-154:2250247:2250742 [0] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2250247:2250742 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2250247:2250742 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2250247:2250742 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2250247:2250742 [0] NCCL INFO Using network IB n136-128-154:2250248:2250743 [1] NCCL INFO ncclCommInitRankConfig comm 0x14b39720 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x6c6b8b0590fd8c7a - Init START n136-128-154:2250247:2250742 [0] NCCL INFO ncclCommInitRankConfig comm 0x167514c0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x6c6b8b0590fd8c7a - Init START n136-128-154:2250247:2250742 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2250248:2250743 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2250248:2250743 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2250248:2250743 [1] NCCL INFO Retrieving state for IB n136-128-154:2250248:2250743 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2250248:2250743 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2250247:2250742 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2250247:2250742 [0] NCCL INFO Retrieving state for IB n136-128-154:2250247:2250742 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2250248:2250743 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2250247:2250742 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2250248:2250743 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2250247:2250742 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2250248:2250743 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2250247:2250742 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2250247:2250742 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2250248:2250743 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250248:2250743 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250247:2250742 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250247:2250742 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250248:2250743 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2250248:2250743 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2250248:2250743 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2250248:2250743 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2250248:2250743 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2250248:2250743 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2250248:2250743 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2250248:2250743 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2250248:2250743 [1] NCCL INFO ========================================== n136-128-154:2250248:2250743 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250248:2250743 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250248:2250743 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2250247:2250742 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2250247:2250742 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2250247:2250742 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2250247:2250742 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2250247:2250742 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2250247:2250742 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2250247:2250742 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2250247:2250742 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2250247:2250742 [0] NCCL INFO ========================================== n136-128-154:2250247:2250742 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250247:2250742 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250247:2250742 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2250247:2250742 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2250247:2250742 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2250247:2250742 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250247:2250742 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250247:2250742 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250247:2250742 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250247:2250742 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250247:2250742 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250247:2250742 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2250248:2250743 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2250248:2250743 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250248:2250743 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250248:2250743 [1] NCCL INFO comm 0x14b39720 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO comm 0x167514c0 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250247:2250742 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2250248:2250743 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2250247:2250742 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2250247:2250742 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2250247:2250742 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2250247:2250742 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2250247:2250742 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2250247:2250755 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 104 n136-128-154:2250247:2250754 [0] NCCL INFO [Proxy Service] Device 0 CPU core 39 n136-128-154:2250248:2250743 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2250248:2250756 [1] NCCL INFO [Proxy Service] Device 1 CPU core 34 n136-128-154:2250248:2250757 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 41 n136-128-154:2250247:2250742 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2250247:2250742 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2250248:2250743 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2250248:2250743 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2250247:2250742 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2250247:2250742 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2250247:2250742 [0] NCCL INFO ncclCommInitRankConfig comm 0x167514c0 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x6c6b8b0590fd8c7a - Init COMPLETE n136-128-154:2250247:2250742 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.52 (kernels 0.12, alloc 0.19, bootstrap 0.00, allgathers 0.01, topo 0.08, graphs 0.00, connections 0.06, rest 0.06) n136-128-154:2250248:2250743 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2250248:2250743 [1] NCCL INFO ncclCommInitRankConfig comm 0x14b39720 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x6c6b8b0590fd8c7a - Init COMPLETE n136-128-154:2250248:2250743 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.51 (kernels 0.12, alloc 0.17, bootstrap 0.00, allgathers 0.01, topo 0.08, graphs 0.00, connections 0.06, rest 0.06) n136-128-154:2250247:2250758 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250759 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250247:2250758 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2250248:2250759 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:20:08:28 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:20:08:28 INFO [evaluator:559] Running loglikelihood requests Passed argument batch_size = auto:1. Detecting largest batch size Running loglikelihood requests: 0%| | 0/1838 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250248:2250834 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2250248:2250839 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250248:2250839 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250248:2250839 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250248:2250839 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250248:2250839 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250248:2250839 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250248:2250756 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:20:08:58 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: auto (64) |Tasks|Version|Filter|n-shot| Metric | |Value | |Stderr| |-----|------:|------|-----:|--------|---|-----:|---|-----:| |piqa | 1|none | 0|acc |↑ |0.6431|± |0.0112| | | |none | 0|acc_norm|↑ |0.6458|± |0.0112| [rank0]:[W1209 20:08:59.922555932 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2250247:2250886 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250247:2250886 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250247:2250754 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2250247:2250886 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250247:2250886 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250247:2250886 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250247:2250886 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250248:2250756 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2250248:2250839 [1] NCCL INFO comm 0x14b39720 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2250247:2250886 [0] NCCL INFO comm 0x167514c0 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 piqa 评估完成! ================================================== 开始评估:任务=hellaswag | 少样本数=10 | 模型=Llama-3.1-8B-quantization-fg 输出路径:results2/Llama-3.1-8B-quantization-fg/fg6/hellaswag.json ================================================== The following values were not passed to `accelerate launch` and had defaults used instead: More than one GPU was found, enabling multi-GPU training. If this was unintended please pass in `--num_processes=1`. `--num_machines` was set to a value of `1` `--mixed_precision` was set to a value of `'no'` `--dynamo_backend` was set to a value of `'no'` To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`. 2025-12-09:20:09:49 INFO [__main__:440] Selected Tasks: ['hellaswag'] 2025-12-09:20:09:49 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:09:49 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:09:49 INFO [__main__:440] Selected Tasks: ['hellaswag'] 2025-12-09:20:09:49 INFO [evaluator:189] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2025-12-09:20:09:49 INFO [evaluator:227] Initializing hf model, with arguments: {'pretrained': '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg'} 2025-12-09:20:09:50 WARNING [accelerate.utils.other:513] Detected kernel version 5.4.143, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher. The tokenizer you are loading from '/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue. 2025-12-09:20:09:51 INFO [models.huggingface:382] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'} `torch_dtype` is deprecated! Use `dtype` instead! Loading checkpoint shards: 0%| | 0/4 [00:00 n136-128-154:2250936:2251207 [1] NCCL INFO Initialized NET plugin IB n136-128-154:2250936:2251207 [1] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2250936:2251207 [1] NCCL INFO Using network IB n136-128-154:2250935:2251206 [0] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1. n136-128-154:2250935:2251206 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [1]mlx5_1:1/RoCE [2]mlx5_2:1/RoCE [3]mlx5_3:1/RoCE [RO]; OOB eth0:fdbd:dc03:9:451::154<0> n136-128-154:2250935:2251206 [0] NCCL INFO Initialized NET plugin IB n136-128-154:2250935:2251206 [0] NCCL INFO Assigned NET plugin IB to comm n136-128-154:2250935:2251206 [0] NCCL INFO Using network IB n136-128-154:2250936:2251207 [1] NCCL INFO ncclCommInitRankConfig comm 0x13dbbd10 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x28bdcd3d8c23a13b - Init START n136-128-154:2250935:2251206 [0] NCCL INFO ncclCommInitRankConfig comm 0x140bf390 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x28bdcd3d8c23a13b - Init START n136-128-154:2250936:2251207 [1] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2250935:2251206 [0] NCCL INFO RAS client listening socket at ::1<28028> n136-128-154:2250936:2251207 [1] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2250936:2251207 [1] NCCL INFO Retrieving state for IB n136-128-154:2250936:2251207 [1] NCCL INFO Initialized state 0 for IB n136-128-154:2250936:2251207 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2250935:2251206 [0] NCCL INFO TOPO/NET : Importing network plugins to topology n136-128-154:2250935:2251206 [0] NCCL INFO Retrieving state for IB n136-128-154:2250935:2251206 [0] NCCL INFO Initialized state 0 for IB n136-128-154:2250936:2251207 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2250935:2251206 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_0 in topo with pciPath=/sys/devices/pci0000:09/0000:09:02.0/0000:0a:00.0/0000:0b:08.0/0000:1b:00.0/0000:1c:00.0/0000:1d:00.0 keep=1 coll=(null) n136-128-154:2250935:2251206 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_1 in topo with pciPath=/sys/devices/pci0000:43/0000:43:02.0/0000:44:00.0/0000:45:08.0/0000:5e:00.0/0000:5f:00.0/0000:60:00.0 keep=1 coll=(null) n136-128-154:2250936:2251207 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2250936:2251207 [1] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2250935:2251206 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_2 in topo with pciPath=/sys/devices/pci0000:82/0000:82:02.0/0000:83:00.0/0000:84:08.0/0000:93:00.0/0000:94:00.0/0000:95:00.0 keep=1 coll=(null) n136-128-154:2250935:2251206 [0] NCCL INFO ncclTopoPopulateNics : Filled mlx5_3 in topo with pciPath=/sys/devices/pci0000:be/0000:be:02.0/0000:bf:00.0/0000:c0:04.0/0000:ca:00.0/0000:cb:10.0/0000:cc:00.0 keep=1 coll=(null) n136-128-154:2250935:2251206 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250935:2251206 [0] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250936:2251207 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 0 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250936:2251207 [1] NCCL INFO GPU Direct RDMA Enabled for GPU 1 / HCA 3 (distance 5 <= 5), read 0 mode Default n136-128-154:2250936:2251207 [1] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2250936:2251207 [1] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2250936:2251207 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2250936:2251207 [1] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2250936:2251207 [1] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2250936:2251207 [1] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2250936:2251207 [1] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2250936:2251207 [1] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2250936:2251207 [1] NCCL INFO ========================================== n136-128-154:2250936:2251207 [1] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250936:2251207 [1] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250936:2251207 [1] NCCL INFO Setting affinity for GPU 7 to 33-62,97-126 n136-128-154:2250935:2251206 [0] NCCL INFO === System : maxBw 240.0 totalBw 240.0 === n136-128-154:2250935:2251206 [0] NCCL INFO CPU/0-1 (1/1/2) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - PCI/0-83000 (1000c0101000ffff) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - NIC/0-95000 n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - PCI/0-bf000 (1000c0101000ffff) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - PCI/0-c3000 (1000c01010de13b8) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - GPU/0-c5000 (0) n136-128-154:2250935:2251206 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - PCI/0-c7000 (1000c01010de13b8) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - GPU/0-c9000 (1) n136-128-154:2250935:2251206 [0] NCCL INFO + NVL[240.0] - NVS/0-0 n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - NIC/0-cc000 n136-128-154:2250935:2251206 [0] NCCL INFO + SYS[10.0] - CPU/0-0 n136-128-154:2250935:2251206 [0] NCCL INFO CPU/0-0 (1/1/2) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - PCI/0-a000 (1000c0101000ffff) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - NIC/0-1d000 n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - PCI/0-44000 (1000c0101000ffff) n136-128-154:2250935:2251206 [0] NCCL INFO + PCI[24.0] - NIC/0-60000 n136-128-154:2250935:2251206 [0] NCCL INFO + SYS[10.0] - CPU/0-1 n136-128-154:2250935:2251206 [0] NCCL INFO ========================================== n136-128-154:2250935:2251206 [0] NCCL INFO GPU/0-c5000 :GPU/0-c5000 (0/5000.0/LOC) GPU/0-c9000 (2/240.0/NVL) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250935:2251206 [0] NCCL INFO GPU/0-c9000 :GPU/0-c5000 (2/240.0/NVL) GPU/0-c9000 (0/5000.0/LOC) NVS/0-0 (1/240.0/NVL) CPU/0-1 (3/24.0/PHB) CPU/0-0 (4/10.0/SYS) n136-128-154:2250935:2251206 [0] NCCL INFO Setting affinity for GPU 6 to 33-62,97-126 n136-128-154:2250936:2251207 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2250936:2251207 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2250936:2251207 [1] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250936:2251207 [1] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250936:2251207 [1] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250936:2251207 [1] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250936:2251207 [1] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250936:2251207 [1] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250936:2251207 [1] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250935:2251206 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 20.000000/20.000000, type NVL/PIX, sameChannels 1 n136-128-154:2250935:2251206 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 6 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 7 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 8 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 9 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 10 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 11 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 40.000000/40.000000, type NVL/PIX, sameChannels 0 n136-128-154:2250935:2251206 [0] NCCL INFO 0 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 1 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 2 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 3 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 4 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 5 : GPU/0-c5000 GPU/0-c9000 n136-128-154:2250935:2251206 [0] NCCL INFO 6 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250935:2251206 [0] NCCL INFO 7 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250935:2251206 [0] NCCL INFO 8 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250935:2251206 [0] NCCL INFO 9 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250935:2251206 [0] NCCL INFO 10 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250935:2251206 [0] NCCL INFO 11 : GPU/0-c9000 GPU/0-c5000 n136-128-154:2250936:2251207 [1] NCCL INFO comm 0x13dbbd10 rank 1 nRanks 2 nNodes 1 localRanks 2 localRank 1 MNNVL 0 n136-128-154:2250935:2251206 [0] NCCL INFO comm 0x140bf390 rank 0 nRanks 2 nNodes 1 localRanks 2 localRank 0 MNNVL 0 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 0 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 12 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 1 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 13 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 2 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 14 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 3 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 15 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 4 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 16 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 5 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 17 : 0 -> 1 -> -1/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 6 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 6 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 18 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 18 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 7 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 7 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 19 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 19 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 8 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 8 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 20 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 20 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 9 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 9 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 21 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 21 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 10 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 10 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 22 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 22 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 11 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 11 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Tree 23 : 1 -> 0 -> -1/-1/-1 n136-128-154:2250936:2251207 [1] NCCL INFO Tree 23 : -1 -> 1 -> 0/-1/-1 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 00/24 : 0 1 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 01/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 00 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 02/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 01 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 03/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 02 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 04/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 03 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 05/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 04 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 06/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 05 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 07/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 06 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 08/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 07 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 09/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 08 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 10/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 09 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 11/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 10 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 12/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 11 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 13/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 12 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 14/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 13 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 15/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 14 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 16/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 15 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 17/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 16 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 18/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 17 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 19/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 18 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 20/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 19 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 21/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 20 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 22/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 21 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Channel 23/24 : 0 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 22 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 00 : 1 -> 0 -> 1 n136-128-154:2250936:2251207 [1] NCCL INFO Ring 23 : 0 -> 1 -> 0 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 01 : 1 -> 0 -> 1 n136-128-154:2250936:2251207 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 02 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 03 : 1 -> 0 -> 1 n136-128-154:2250936:2251207 [1] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 04 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 05 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 06 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 07 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 08 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 09 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 10 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 11 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 12 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 13 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 14 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 15 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 16 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 17 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 18 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 19 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 20 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 21 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 22 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Ring 23 : 1 -> 0 -> 1 n136-128-154:2250935:2251206 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1 n136-128-154:2250935:2251206 [0] NCCL INFO P2P Chunksize set to 524288 n136-128-154:2250936:2251207 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2250936:2251220 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 102 n136-128-154:2250936:2251219 [1] NCCL INFO [Proxy Service] Device 1 CPU core 99 n136-128-154:2250935:2251206 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. n136-128-154:2250935:2251206 [0] NCCL INFO Check P2P Type isAllDirectP2p 1 directMode 0 n136-128-154:2250935:2251221 [0] NCCL INFO [Proxy Service] Device 0 CPU core 46 n136-128-154:2250935:2251222 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 47 n136-128-154:2250936:2251207 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2250936:2251207 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2250935:2251206 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512 n136-128-154:2250935:2251206 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer n136-128-154:2250935:2251206 [0] NCCL INFO CC Off, workFifoBytes 1048576 n136-128-154:2250936:2251207 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2250936:2251207 [1] NCCL INFO ncclCommInitRankConfig comm 0x13dbbd10 rank 1 nranks 2 cudaDev 1 nvmlDev 7 busId c9000 commId 0x28bdcd3d8c23a13b - Init COMPLETE n136-128-154:2250936:2251207 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 2 total 0.82 (kernels 0.14, alloc 0.44, bootstrap 0.00, allgathers 0.02, topo 0.08, graphs 0.00, connections 0.01, rest 0.14) n136-128-154:2250935:2251206 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner.so. Using internal tuner plugin. n136-128-154:2250935:2251206 [0] NCCL INFO ncclCommInitRankConfig comm 0x140bf390 rank 0 nranks 2 cudaDev 0 nvmlDev 6 busId c5000 commId 0x28bdcd3d8c23a13b - Init COMPLETE n136-128-154:2250935:2251206 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 2 total 0.83 (kernels 0.14, alloc 0.44, bootstrap 0.00, allgathers 0.02, topo 0.08, graphs 0.00, connections 0.01, rest 0.14) n136-128-154:2250936:2251223 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250935:2251224 [0] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/CUMEM/read n136-128-154:2250936:2251223 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 n136-128-154:2250935:2251224 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 2025-12-09:20:10:44 INFO [evaluator:559] Running loglikelihood requests 2025-12-09:20:10:44 INFO [evaluator:559] Running loglikelihood requests Passed argument batch_size = auto:1. Detecting largest batch size Running loglikelihood requests: 0%| | 0/20084 [00:00 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 01/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 02/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 03/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 04/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 05/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 06/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 07/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 08/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 09/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 10/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 11/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 12/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 13/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 14/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 15/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 16/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 17/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 18/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 19/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 20/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 21/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 22/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 23/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 24/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 25/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 26/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 27/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 28/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 29/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 30/1 : 1[7] -> 0[6] via P2P/CUMEM/read n136-128-154:2250936:2253810 [1] NCCL INFO Channel 31/1 : 1[7] -> 0[6] via P2P/CUMEM/read fatal: detected dubious ownership in repository at '/mnt/bn/life-mllm/users/cxr/quantization' To add an exception for this directory, call: git config --global --add safe.directory /mnt/bn/life-mllm/users/cxr/quantization n136-128-154:2250936:2253816 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250936:2253816 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250936:2253816 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250936:2253816 [1] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250936:2253816 [1] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250936:2253816 [1] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250936:2251219 [1] NCCL INFO misc/socket.cc:915 -> 3 2025-12-09:20:36:05 INFO [loggers.evaluation_tracker:209] Saving results aggregated hf (pretrained=/mnt/bn/life-mllm/users/cxr/quantization/models/Llama-3.1-8B-quantization-fg), gen_kwargs: (None), limit: None, num_fewshot: 10, batch_size: auto (64) | Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr| |---------|------:|------|-----:|--------|---|-----:|---|-----:| |hellaswag| 1|none | 10|acc |↑ |0.3834|± |0.0049| | | |none | 10|acc_norm|↑ |0.5266|± |0.0050| [rank0]:[W1209 20:36:06.916712871 ProcessGroupNCCL.cpp:1524] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) n136-128-154:2250935:2253865 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250935:2253865 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250935:2253865 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250935:2251221 [0] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2250935:2253865 [0] NCCL INFO misc/socket.cc:64 -> 3 n136-128-154:2250935:2253865 [0] NCCL INFO misc/socket.cc:81 -> 3 n136-128-154:2250935:2253865 [0] NCCL INFO misc/socket.cc:863 -> 3 n136-128-154:2250936:2251219 [1] NCCL INFO misc/socket.cc:915 -> 3 n136-128-154:2250936:2253816 [1] NCCL INFO comm 0x13dbbd10 rank 1 nranks 2 cudaDev 1 busId c9000 - Abort COMPLETE n136-128-154:2250935:2253865 [0] NCCL INFO comm 0x140bf390 rank 0 nranks 2 cudaDev 0 busId c5000 - Abort COMPLETE 任务 hellaswag 评估完成! Execution time: 2748.52 seconds