Upload IterCAD-RL

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

args.json

@@ -0,0 +1,532 @@
+{
+  "output_dir": "/mnt/phwfile/datafrontier/hutao/code/ms-swift/output/CAD-RL-GVPM-mixed-sft4500/v2-20260520-115654",
+  "per_device_train_batch_size": 2,
+  "num_train_epochs": 2.0,
+  "max_steps": -1,
+  "learning_rate": 1e-06,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_steps": 0,
+  "optim": "adamw_torch",
+  "optim_args": null,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "optim_target_modules": null,
+  "gradient_accumulation_steps": 4,
+  "average_tokens_across_devices": true,
+  "max_grad_norm": 1.0,
+  "label_smoothing_factor": 0.0,
+  "bf16": true,
+  "fp16": false,
+  "bf16_full_eval": false,
+  "fp16_full_eval": false,
+  "tf32": null,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": null,
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "use_liger_kernel": false,
+  "liger_kernel_config": null,
+  "use_cache": false,
+  "neftune_noise_alpha": null,
+  "torch_empty_cache_steps": null,
+  "auto_find_batch_size": false,
+  "logging_strategy": "steps",
+  "logging_steps": 1,
+  "logging_first_step": true,
+  "log_on_each_node": true,
+  "logging_nan_inf_filter": true,
+  "include_num_input_tokens_seen": false,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "disable_tqdm": null,
+  "report_to": [
+    "tensorboard"
+  ],
+  "run_name": "/mnt/phwfile/datafrontier/hutao/code/ms-swift/output/CAD-RL-GVPM-mixed-sft4500/v2-20260520-115654",
+  "project": "huggingface",
+  "trackio_space_id": "trackio",
+  "eval_strategy": "no",
+  "eval_steps": 10.0,
+  "eval_delay": 0,
+  "per_device_eval_batch_size": 1,
+  "prediction_loss_only": false,
+  "eval_on_start": false,
+  "eval_do_concat_batches": true,
+  "eval_use_gather_object": false,
+  "eval_accumulation_steps": null,
+  "include_for_metrics": [],
+  "batch_eval_metrics": false,
+  "save_only_model": false,
+  "save_strategy": "steps",
+  "save_steps": 10.0,
+  "save_on_each_node": false,
+  "save_total_limit": 10,
+  "enable_jit_checkpoint": false,
+  "push_to_hub": false,
+  "hub_token": null,
+  "hub_private_repo": null,
+  "hub_model_id": null,
+  "hub_strategy": "every_save",
+  "hub_always_push": false,
+  "hub_revision": null,
+  "load_best_model_at_end": false,
+  "metric_for_best_model": "loss",
+  "greater_is_better": false,
+  "ignore_data_skip": false,
+  "restore_callback_states_from_checkpoint": false,
+  "full_determinism": false,
+  "seed": 42,
+  "data_seed": 42,
+  "use_cpu": false,
+  "accelerator_config": {
+    "dispatch_batches": false
+  },
+  "parallelism_config": null,
+  "dataloader_drop_last": false,
+  "dataloader_num_workers": 8,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": false,
+  "dataloader_prefetch_factor": null,
+  "remove_unused_columns": false,
+  "label_names": null,
+  "train_sampling_strategy": "random",
+  "length_column_name": "length",
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "ddp_backend": null,
+  "ddp_timeout": 18000000,
+  "fsdp": [],
+  "fsdp_config": null,
+  "deepspeed": {
+    "fp16": {
+      "enabled": "auto",
+      "loss_scale": 0,
+      "loss_scale_window": 1000,
+      "initial_scale_power": 16,
+      "hysteresis": 2,
+      "min_loss_scale": 1
+    },
+    "bf16": {
+      "enabled": "auto"
+    },
+    "zero_optimization": {
+      "stage": 3,
+      "offload_optimizer": {
+        "device": "none",
+        "pin_memory": true
+      },
+      "offload_param": {
+        "device": "none",
+        "pin_memory": true
+      },
+      "overlap_comm": false,
+      "contiguous_gradients": true,
+      "sub_group_size": 1000000000.0,
+      "reduce_bucket_size": "auto",
+      "zero_quantized_weights": false,
+      "zero_quantized_gradients": false,
+      "stage3_prefetch_bucket_size": "auto",
+      "stage3_param_persistence_threshold": "auto",
+      "stage3_max_live_parameters": 1000000000.0,
+      "stage3_max_reuse_distance": 1000000000.0,
+      "stage3_gather_16bit_weights_on_model_save": true
+    },
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+  },
+  "debug": null,
+  "skip_memory_metrics": true,
+  "do_train": false,
+  "do_eval": false,
+  "do_predict": false,
+  "resume_from_checkpoint": null,
+  "warmup_ratio": 0.05,
+  "logging_dir": "/mnt/phwfile/datafrontier/hutao/code/ms-swift/output/CAD-RL-GVPM-mixed-sft4500/v2-20260520-115654/runs",
+  "local_rank": 0,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "tuner_backend": "peft",
+  "vit_gradient_checkpointing": null,
+  "router_aux_loss_coef": 0.0,
+  "enable_dft_loss": false,
+  "enable_channel_loss": false,
+  "safe_serialization": true,
+  "max_shard_size": "5GB",
+  "check_model": true,
+  "acc_strategy": "token",
+  "train_dataloader_shuffle": true,
+  "group_by_length": false,
+  "max_epochs": null,
+  "aligner_lr": null,
+  "vit_lr": null,
+  "use_logits_to_keep": null,
+  "ds3_gather_for_generation": true,
+  "resume_only_model": false,
+  "optimizer": null,
+  "loss_type": "grpo",
+  "eval_metric": null,
+  "callbacks": [
+    "cad_eval"
+  ],
+  "early_stop_interval": null,
+  "eval_use_evalscope": false,
+  "eval_dataset": [],
+  "eval_dataset_args": null,
+  "eval_limit": null,
+  "eval_generation_config": null,
+  "extra_eval_args": null,
+  "tuner_type": "full",
+  "use_galore": false,
+  "galore_target_modules": null,
+  "galore_rank": 128,
+  "galore_update_proj_gap": 50,
+  "galore_scale": 1.0,
+  "galore_proj_type": "std",
+  "galore_optim_per_parameter": false,
+  "galore_with_embedding": false,
+  "galore_quantization": false,
+  "galore_proj_quant": false,
+  "galore_proj_bits": 4,
+  "galore_proj_group_size": 256,
+  "galore_cos_threshold": 0.4,
+  "galore_gamma_proj": 2,
+  "galore_queue_size": 5,
+  "lisa_activated_layers": 0,
+  "lisa_step_interval": 20,
+  "use_flash_ckpt": false,
+  "use_ray": false,
+  "ray_exp_name": null,
+  "device_groups": null,
+  "model": "/mnt/phwfile/datafrontier/hutao/code/ms-swift/stage2_clean_thinking_cleanview_add_new_sft_3_fix_format_edit_good/v0-20260513-022358/checkpoint-4500",
+  "model_type": "qwen3_5",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": null,
+  "experts_impl": null,
+  "new_special_tokens": [],
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "max_model_len": null,
+  "local_repo_path": null,
+  "init_strategy": null,
+  "template": "qwen3_5",
+  "system": null,
+  "max_length": 32768,
+  "truncation_strategy": "left",
+  "max_pixels": null,
+  "agent_template": null,
+  "norm_bbox": null,
+  "use_chat_template": true,
+  "padding_side": "right",
+  "padding_free": false,
+  "loss_scale": "default",
+  "sequence_parallel_size": 1,
+  "template_backend": "swift",
+  "response_prefix": null,
+  "enable_thinking": false,
+  "add_non_thinking_prefix": true,
+  "dataset": [
+    "/mnt/phwfile/datafrontier/hutao/code/ms-swift/data/cad_rl_mixed.jsonl"
+  ],
+  "val_dataset": [],
+  "cached_dataset": [],
+  "cached_val_dataset": [],
+  "split_dataset_ratio": 0.0,
+  "dataset_num_proc": 1,
+  "load_from_cache_file": true,
+  "dataset_shuffle": true,
+  "val_dataset_shuffle": false,
+  "streaming": false,
+  "interleave_prob": null,
+  "stopping_strategy": "first_exhausted",
+  "shuffle_buffer_size": 1000,
+  "download_mode": "reuse_dataset_if_exists",
+  "columns": {},
+  "strict": false,
+  "model_name": null,
+  "model_author": null,
+  "custom_dataset_info": [],
+  "quant_method": null,
+  "quant_bits": null,
+  "hqq_axis": null,
+  "bnb_4bit_compute_dtype": "bfloat16",
+  "bnb_4bit_quant_type": "nf4",
+  "bnb_4bit_use_double_quant": true,
+  "bnb_4bit_quant_storage": null,
+  "max_new_tokens": 8192,
+  "temperature": 1.0,
+  "top_k": 50,
+  "top_p": 0.9,
+  "repetition_penalty": 1.0,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "structured_outputs_regex": null,
+  "train_type": "full",
+  "adapters": [],
+  "external_plugins": [
+    "/mnt/phwfile/datafrontier/hutao/code/ms-swift/examples/train/grpo/plugin/cad_grpo_plugin.py"
+  ],
+  "custom_register_path": [],
+  "model_kwargs": {},
+  "load_args": false,
+  "load_data_args": false,
+  "packing": false,
+  "packing_length": null,
+  "packing_num_proc": 1,
+  "lazy_tokenize": true,
+  "use_hf": false,
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "freeze_parameters": [
+    "model.visual",
+    "model.visual.merger"
+  ],
+  "freeze_parameters_regex": null,
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [],
+  "trainable_parameters_regex": null,
+  "freeze_llm": false,
+  "freeze_vit": true,
+  "freeze_aligner": true,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "target_parameters": null,
+  "modules_to_save": [],
+  "lora_rank": 8,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": false,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": "ms-swift",
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_notification_method": null,
+  "swanlab_webhook_url": null,
+  "swanlab_secret": null,
+  "swanlab_sender_email": null,
+  "swanlab_receiver_email": null,
+  "swanlab_smtp_server": null,
+  "swanlab_smtp_port": null,
+  "swanlab_email_language": "zh",
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "create_checkpoint_symlink": false,
+  "zero_hpz_partition_size": null,
+  "deepspeed_autotp_size": null,
+  "reward_model": null,
+  "reward_adapters": [],
+  "reward_model_type": null,
+  "reward_model_revision": null,
+  "num_ppo_epochs": 4,
+  "whiten_rewards": false,
+  "kl_coef": 0.05,
+  "cliprange": 0.2,
+  "vf_coef": 0.1,
+  "cliprange_value": 0.2,
+  "gamma": 1.0,
+  "lam": 0.95,
+  "num_mini_batches": 1,
+  "local_rollout_forward_batch_size": 64,
+  "num_sample_generations": 10,
+  "response_length": 8192,
+  "missing_eos_penalty": null,
+  "vllm_gpu_memory_utilization": 0.5,
+  "vllm_tensor_parallel_size": 1,
+  "vllm_pipeline_parallel_size": 1,
+  "vllm_enable_expert_parallel": false,
+  "vllm_max_num_seqs": null,
+  "vllm_max_model_len": 32768,
+  "vllm_disable_custom_all_reduce": true,
+  "vllm_enforce_eager": false,
+  "vllm_limit_mm_per_prompt": null,
+  "vllm_max_lora_rank": 16,
+  "vllm_enable_prefix_caching": true,
+  "vllm_use_async_engine": null,
+  "vllm_quantization": null,
+  "vllm_reasoning_parser": null,
+  "vllm_disable_cascade_attn": false,
+  "vllm_mm_processor_cache_gb": 0.0,
+  "vllm_speculative_config": null,
+  "vllm_engine_kwargs": {},
+  "vllm_data_parallel_size": 1,
+  "use_vllm": true,
+  "vllm_mode": "colocate",
+  "vllm_enable_lora": false,
+  "vllm_server_base_url": null,
+  "vllm_server_host": null,
+  "vllm_server_port": [
+    8000
+  ],
+  "vllm_server_timeout": 240.0,
+  "vllm_server_group_port": null,
+  "enable_flattened_weight_sync": true,
+  "async_generate": false,
+  "sleep_level": 1,
+  "move_model_batches": null,
+  "offload_optimizer": true,
+  "offload_model": true,
+  "wandb_log_unique_prompts": null,
+  "epsilon": 0.0003,
+  "epsilon_high": 0.0004,
+  "delta": null,
+  "cosine_min_len_value_wrong": -0.5,
+  "cosine_max_len_value_wrong": 0.0,
+  "cosine_min_len_value_correct": 1.0,
+  "cosine_max_len_value_correct": 0.5,
+  "cosine_max_len": null,
+  "repetition_n_grams": 3,
+  "repetition_max_penalty": -1.0,
+  "reward_model_plugin": null,
+  "chord_sft_dataset": [],
+  "chord_sft_per_device_train_batch_size": null,
+  "chord_enable_phi_function": false,
+  "chord_mu_warmup_steps": null,
+  "chord_mu_decay_steps": null,
+  "chord_mu_peak": null,
+  "chord_mu_valley": null,
+  "sync_ref_model": false,
+  "ref_model_sync_steps": 512,
+  "ref_model_mixup_alpha": 0.6,
+  "multi_turn_scheduler": "cad_scheduler",
+  "max_turns": 5,
+  "completion_length_limit_scope": "per_round",
+  "vllm_server_pass_dataset": false,
+  "dynamic_sample": false,
+  "max_resample_times": 3,
+  "overlong_filter": false,
+  "soft_max_length": null,
+  "soft_cache_length": null,
+  "scale_rewards": "none",
+  "log_entropy": false,
+  "top_entropy_quantile": 1.0,
+  "importance_sampling_level": "sequence",
+  "tau_pos": 1.0,
+  "tau_neg": 1.05,
+  "advantage_estimator": "grpo",
+  "kl_in_reward": false,
+  "generation_batch_size": null,
+  "steps_per_generation": 4,
+  "num_generations_eval": null,
+  "rollout_importance_sampling_mode": null,
+  "rollout_importance_sampling_threshold": 2.0,
+  "log_rollout_offpolicy_metrics": false,
+  "off_policy_sequence_mask_delta": null,
+  "num_generations": 8,
+  "reward_funcs": [
+    "cad_chamfer",
+    "cad_format",
+    "cad_progress",
+    "cad_cd_value",
+    "cad_invalid"
+  ],
+  "reward_weights": [
+    1.0,
+    0.5,
+    0.5,
+    0.0,
+    0.0
+  ],
+  "log_completions": true,
+  "num_iterations": 1,
+  "teacher_model": null,
+  "teacher_adapters": [],
+  "teacher_model_type": null,
+  "teacher_model_revision": null,
+  "teacher_deepspeed": null,
+  "teacher_model_server": null,
+  "rlhf_type": "grpo",
+  "ref_model": null,
+  "ref_adapters": [],
+  "ref_model_type": null,
+  "ref_model_revision": null,
+  "beta": 0.0,
+  "label_smoothing": 0,
+  "max_completion_length": 8192,
+  "rpo_alpha": null,
+  "ld_alpha": null,
+  "discopop_tau": 0.05,
+  "loss_weights": null,
+  "cpo_alpha": 1.0,
+  "simpo_gamma": 1,
+  "desirable_weight": 1.0,
+  "undesirable_weight": 1.0,
+  "center_rewards_coefficient": null,
+  "sft_alpha": 0,
+  "lmbda": 0.5,
+  "seq_kd": false,
+  "gkd_logits_topk": null,
+  "offload_teacher_model": false,
+  "swift_version": "4.1.0.dev0",
+  "ckpt_dir": "/mnt/phwfile/datafrontier/hutao/code/ms-swift/stage2_clean_thinking_cleanview_add_new_sft_3_fix_format_edit_good/v0-20260513-022358/checkpoint-4500",
+  "rank": 0,
+  "global_world_size": 8,
+  "local_world_size": 8,
+  "model_suffix": "checkpoint-4500",
+  "model_info": "ModelInfo(model_type='qwen3_5', model_dir='/mnt/phwfile/datafrontier/hutao/code/ms-swift/stage2_clean_thinking_cleanview_add_new_sft_3_fix_format_edit_good/v0-20260513-022358/checkpoint-4500', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
+  "model_meta": "ModelMeta(model_type='qwen3_5', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3.5-0.8B', hf_model_id='Qwen/Qwen3.5-0.8B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-2B', hf_model_id='Qwen/Qwen3.5-2B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-4B', hf_model_id='Qwen/Qwen3.5-4B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-9B', hf_model_id='Qwen/Qwen3.5-9B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-27B', hf_model_id='Qwen/Qwen3.5-27B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-27B-FP8', hf_model_id='Qwen/Qwen3.5-27B-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-0.8B-Base', hf_model_id='Qwen/Qwen3.5-0.8B-Base', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-2B-Base', hf_model_id='Qwen/Qwen3.5-2B-Base', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-4B-Base', hf_model_id='Qwen/Qwen3.5-4B-Base', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-9B-Base', hf_model_id='Qwen/Qwen3.5-9B-Base', model_path=None, ms_revision=None, hf_revision=None)], template='qwen3_5', ignore_patterns=None, requires=None, tags=[])], loader=<class 'swift.model.models.qwen.Qwen3_5Loader'>, template=None, model_arch=MultiModelKeys(arch_name='qwen2_vl', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp=None, down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model', 'lm_head'], aligner=['model.visual.merger'], vision_tower=['model.visual'], generator=[]), architectures=['Qwen3_5ForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=5.0.0.dev', 'qwen_vl_utils>=0.0.14', 'decord'], tags=['vision', 'video'])",
+  "model_dir": "/mnt/phwfile/datafrontier/hutao/code/ms-swift/stage2_clean_thinking_cleanview_add_new_sft_3_fix_format_edit_good/v0-20260513-022358/checkpoint-4500",
+  "template_meta": "QwenTemplateMeta(template_type='qwen3_5', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'swift.template.templates.qwen.Qwen3_5Template'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='qwen3_5', is_thinking=True, thinking_prefix='<think>\\n', non_thinking_prefix='<think>\\n\\n</think>\\n\\n', history_thinking_prefix='')",
+  "_val_dataset_exists": false,
+  "hub": "<class 'swift.hub.hub.MSHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "GRPOConfig(output_dir='/mnt/phwfile/datafrontier/hutao/code/ms-swift/output/CAD-RL-GVPM-mixed-sft4500/v2-20260520-115654', per_device_train_batch_size=2, num_train_epochs=2.0, max_steps=-1, learning_rate=1e-06, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=0.05, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=4, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=False, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=1, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=['tensorboard'], run_name='/mnt/phwfile/datafrontier/hutao/code/ms-swift/output/CAD-RL-GVPM-mixed-sft4500/v2-20260520-115654', project='huggingface', trackio_space_id='trackio', eval_strategy=<IntervalStrategy.NO: 'no'>, eval_steps=10.0, eval_delay=0, per_device_eval_batch_size=1, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=10, save_on_each_node=False, save_total_limit=10, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=True, dataloader_num_workers=8, dataloader_pin_memory=True, dataloader_persistent_workers=False, dataloader_prefetch_factor=2, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_backend=None, ddp_timeout=18000000, fsdp=[], fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'none', 'pin_memory': True}, 'overlap_comm': False, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'zero_quantized_weights': False, 'zero_quantized_gradients': False, 'stage3_prefetch_bucket_size': 0, 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, debug=[], skip_memory_metrics=True, do_train=False, do_eval=False, do_predict=False, resume_from_checkpoint=None, warmup_ratio=0.05, logging_dir='/mnt/phwfile/datafrontier/hutao/code/ms-swift/output/CAD-RL-GVPM-mixed-sft4500/v2-20260520-115654/runs', local_rank=0, model_init_kwargs=None, disable_dropout=False, cast_lm_head_to_fp32=False, num_generations=8, num_generations_eval=None, max_completion_length=8192, ds3_gather_for_generation=True, shuffle_dataset=True, generation_batch_size=64, steps_per_generation=4, temperature=1.0, top_p=0.9, top_k=50, min_p=None, generation_kwargs=None, chat_template_kwargs=None, repetition_penalty=1.0, use_transformers_paged=False, cache_implementation=None, use_vllm=True, vllm_mode='colocate', vllm_model_impl='vllm', vllm_enable_sleep_mode=False, vllm_structured_outputs_regex=None, vllm_server_base_url=None, vllm_server_host=None, vllm_server_port=[8000], vllm_server_timeout=240.0, vllm_group_port=51216, vllm_gpu_memory_utilization=0.5, vllm_max_model_length=None, vllm_tensor_parallel_size=1, beta=0.0, num_iterations=1, epsilon=0.0003, delta=None, epsilon_high=0.0004, sapo_temperature_neg=1.05, sapo_temperature_pos=1.0, importance_sampling_level='sequence', reward_weights=[1.0, 0.5, 0.5, 0.0, 0.0], multi_objective_aggregation='sum_then_normalize', scale_rewards='none', loss_type='grpo', mask_truncated_completions=False, sync_ref_model=False, ref_model_mixup_alpha=0.6, ref_model_sync_steps=512, top_entropy_quantile=1.0, max_tool_calling_iterations=None, vllm_importance_sampling_correction=True, vllm_importance_sampling_mode='sequence_mask', vllm_importance_sampling_cap=3.0, off_policy_mask_threshold=None, use_bias_correction_kl=False, log_completions=True, num_completions_to_print=None, log_unique_prompts=False, log_completions_hub_repo=None, tuner_backend='peft', vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=True, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, max_epochs=None, aligner_lr=None, vit_lr=None, use_logits_to_keep=None, resume_only_model=False, optimizer=None, eval_metric=None, callbacks=['cad_eval'], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='full', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False, vllm_pipeline_parallel_size=1, vllm_enable_expert_parallel=False, vllm_max_num_seqs=None, vllm_max_model_len=32768, vllm_disable_custom_all_reduce=True, vllm_enforce_eager=False, vllm_limit_mm_per_prompt=None, vllm_max_lora_rank=16, vllm_enable_prefix_caching=True, vllm_use_async_engine=None, vllm_quantization=None, vllm_reasoning_parser=None, vllm_disable_cascade_attn=False, vllm_mm_processor_cache_gb=0.0, vllm_speculative_config=None, vllm_engine_kwargs={}, vllm_data_parallel_size=1, stop_words=[], vllm_enable_lora=False, lora_rank=8, vllm_server_group_port=None, enable_flattened_weight_sync=True, async_generate=False, structured_outputs_regex=None, sleep_level=1, move_model_batches=None, offload_optimizer=True, offload_model=True, wandb_log_unique_prompts=None, cosine_min_len_value_wrong=-0.5, cosine_max_len_value_wrong=0.0, cosine_min_len_value_correct=1.0, cosine_max_len_value_correct=0.5, cosine_max_len=8192, repetition_n_grams=3, repetition_max_penalty=-1.0, reward_model=None, reward_model_plugin=None, chord_sft_dataset=[], chord_sft_per_device_train_batch_size=None, chord_enable_phi_function=False, chord_mu_warmup_steps=None, chord_mu_decay_steps=None, chord_mu_peak=None, chord_mu_valley=None, multi_turn_scheduler='cad_scheduler', max_turns=5, completion_length_limit_scope='per_round', vllm_server_pass_dataset=False, dynamic_sample=False, max_resample_times=3, overlong_filter=False, soft_max_length=None, soft_cache_length=None, log_entropy=False, tau_pos=1.0, tau_neg=1.05, advantage_estimator='grpo', kl_in_reward=False, dataset_shuffle=True, rollout_importance_sampling_mode=None, rollout_importance_sampling_threshold=2.0, log_rollout_offpolicy_metrics=False, off_policy_sequence_mask_delta=None)"
+}

chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

config.json

@@ -0,0 +1,113 @@
+{
+  "architectures": [
+    "Qwen3_5ForConditionalGeneration"
+  ],
+  "dtype": "bfloat16",
+  "eos_token_id": 248046,
+  "hidden_size": 2560,
+  "image_token_id": 248056,
+  "model_type": "qwen3_5",
+  "pad_token_id": 248044,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_output_gate": true,
+    "bos_token_id": null,
+    "dtype": "bfloat16",
+    "eos_token_id": 248044,
+    "full_attention_interval": 4,
+    "head_dim": 256,
+    "hidden_act": "silu",
+    "hidden_size": 2560,
+    "initializer_range": 0.02,
+    "intermediate_size": 9216,
+    "layer_types": [
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention"
+    ],
+    "linear_conv_kernel_dim": 4,
+    "linear_key_head_dim": 128,
+    "linear_num_key_heads": 16,
+    "linear_num_value_heads": 32,
+    "linear_value_head_dim": 128,
+    "mamba_ssm_dtype": "float32",
+    "max_position_embeddings": 262144,
+    "mlp_only_layers": [],
+    "model_type": "qwen3_5_text",
+    "mtp_num_hidden_layers": 1,
+    "mtp_use_dedicated_embeddings": false,
+    "num_attention_heads": 16,
+    "num_hidden_layers": 32,
+    "num_key_value_heads": 4,
+    "pad_token_id": 248044,
+    "partial_rotary_factor": 0.25,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        11,
+        11,
+        10
+      ],
+      "partial_rotary_factor": 0.25,
+      "rope_theta": 10000000,
+      "rope_type": "default"
+    },
+    "tie_word_embeddings": true,
+    "use_cache": false,
+    "vocab_size": 248320
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.2.0",
+  "use_cache": false,
+  "video_token_id": 248057,
+  "vision_config": {
+    "deepstack_visual_indexes": [],
+    "depth": 24,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1024,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 4096,
+    "model_type": "qwen3_5",
+    "num_heads": 16,
+    "num_position_embeddings": 2304,
+    "out_hidden_size": 2560,
+    "patch_size": 16,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2
+  },
+  "vision_end_token_id": 248054,
+  "vision_start_token_id": 248053
+}

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    248044,
+    248046
+  ],
+  "transformers_version": "5.2.0",
+  "use_cache": true
+}

latest

@@ -0,0 +1 @@
+global_step70

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdfb4ea9d384e40e4738845ce32ac2d7caa112a891d15319b91400d1cef17ad8
+size 4958192864

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3904f7bd6b3da3a75947841e50b4c021e807d5563655550cde3480794e62fde
+size 4993498424

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5fe504876af0c6a7e8f604bd1ce04f758feddb20ae0675c6252cb27752ed8a47
+size 398327248

model.safetensors.index.json

@@ -0,0 +1,732 @@
+{
+  "metadata": {
+    "total_parameters": 504320,
+    "total_size": 10349929472
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.embed_tokens.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.13.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.14.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.15.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.16.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.17.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.18.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.19.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.2.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.20.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.20.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.21.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.22.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.23.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.24.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.25.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.26.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.27.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.28.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.29.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.3.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.3.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.30.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.A_log": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.dt_bias": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.30.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.31.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.language_model.layers.4.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.4.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.5.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.6.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.7.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.8.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.A_log": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.dt_bias": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.norm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.layers.9.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.language_model.norm.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.0.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.1.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.10.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.10.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.10.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.10.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.11.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.12.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.13.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.14.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.15.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.16.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.17.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.18.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.19.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.2.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.2.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.20.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.20.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.21.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.22.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.attn.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.attn.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.attn.qkv.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.attn.qkv.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.norm1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.norm1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.norm2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.23.norm2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.blocks.3.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.3.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.4.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.5.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.6.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.7.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.8.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.attn.proj.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.attn.proj.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.attn.qkv.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.attn.qkv.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.norm1.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.norm1.weight": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.norm2.bias": "model-00002-of-00003.safetensors",
+    "model.visual.blocks.9.norm2.weight": "model-00002-of-00003.safetensors",
+    "model.visual.merger.linear_fc1.bias": "model-00003-of-00003.safetensors",
+    "model.visual.merger.linear_fc1.weight": "model-00003-of-00003.safetensors",
+    "model.visual.merger.linear_fc2.bias": "model-00003-of-00003.safetensors",
+    "model.visual.merger.linear_fc2.weight": "model-00003-of-00003.safetensors",
+    "model.visual.merger.norm.bias": "model-00003-of-00003.safetensors",
+    "model.visual.merger.norm.weight": "model-00003-of-00003.safetensors",
+    "model.visual.patch_embed.proj.bias": "model-00003-of-00003.safetensors",
+    "model.visual.patch_embed.proj.weight": "model-00003-of-00003.safetensors",
+    "model.visual.pos_embed.weight": "model-00003-of-00003.safetensors"
+  }
+}

preprocessor_config.json

@@ -0,0 +1,21 @@
+{
+    "size": {
+        "longest_edge": 16777216,
+        "shortest_edge": 65536
+    },
+    "patch_size": 16,
+    "temporal_patch_size": 2,
+    "merge_size": 2,
+    "image_mean": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "image_std": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "processor_class": "Qwen3VLProcessor",
+    "image_processor_type": "Qwen2VLImageProcessorFast"
+}

processor_config.json

@@ -0,0 +1,63 @@
+{
+  "image_processor": {
+    "data_format": "channels_first",
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Qwen2VLImageProcessorFast",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "merge_size": 2,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "longest_edge": 16777216,
+      "shortest_edge": 65536
+    },
+    "temporal_patch_size": 2
+  },
+  "processor_class": "Qwen3VLProcessor",
+  "video_processor": {
+    "data_format": "channels_first",
+    "default_to_square": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "fps": 2,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_frames": 768,
+    "merge_size": 2,
+    "min_frames": 4,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "longest_edge": 25165824,
+      "shortest_edge": 4096
+    },
+    "temporal_patch_size": 2,
+    "video_processor_type": "Qwen3VLVideoProcessor"
+  }
+}

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc67d7b7557df4207c4a2cd2a68eddd2c330ce612b6702148aaf508ff23f7208
+size 16389

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7528412b251c28f7e14da11381f0fe0dfa8f25c532cfa8337f4ced7ada2e5213
+size 16389

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb379630faf96961c2e575de23d1f092dff768ef9d24192c5865720f84496e03
+size 16389

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ae0b0a025f281cddfacafd8dd06de7a2e4108f355c2a9a437e84a533caf3e68
+size 16389

rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82115f209e4c668628a60f5a7aff11554718e266dd9e548b1902906cab3fb154
+size 16389

rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a01f101fc1312f83fc27ae1a4d37b67137c6716b1546d76d6d2a2bfee3bdd065
+size 16389

rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfa1e9a650fa08e42bb85cd9395e78239c51be3671aa4536cf9a76705eb5732b
+size 16389

rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:354c5b351b369bdfb4c31284ac440f8c329553e29edaa6228c28a270723a06f0
+size 16325

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a539b553fedac07eca23e91eda220efe8225eb38eab8bdb84ec9d6f83275d496
+size 1465

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343

tokenizer_config.json

@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": true,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

trainer_state.json

@@ -0,0 +1,2344 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5185185185185185,
+  "eval_steps": 10.0,
+  "global_step": 70,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.03125,
+      "completions/max_length": 7193.0,
+      "completions/mean_length": 1925.9375,
+      "completions/min_length": 575.0,
+      "epoch": 0.007407407407407408,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.33120759452863396,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 7.142857142857142e-08,
+      "loss": -0.050480857491493225,
+      "num_turns": 3.21875,
+      "reward": 0.9470400810241699,
+      "reward_std": 0.36403971910476685,
+      "rewards/CADCDValueReward/mean": 0.007383444812148809,
+      "rewards/CADCDValueReward/std": 0.013863335829228163,
+      "rewards/CADChamferReward/mean": 0.44704006612300873,
+      "rewards/CADChamferReward/std": 0.3833623081445694,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.12198751419782639,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.03125,
+      "rewards/CADProgressReward/std": 0.0730636678636074,
+      "step": 1,
+      "step_time": 978.6295343909878
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0546875,
+      "completions/max_length": 8451.5,
+      "completions/mean_length": 2523.6796875,
+      "completions/min_length": 638.0,
+      "epoch": 0.014814814814814815,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.25568444681206004,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 1.4285714285714285e-07,
+      "loss": -0.04256395250558853,
+      "num_turns": 3.3828125,
+      "reward": 0.705535352230072,
+      "reward_std": 0.32588590681552887,
+      "rewards/CADCDValueReward/mean": 0.00952278426848352,
+      "rewards/CADCDValueReward/std": 0.012697386788204312,
+      "rewards/CADChamferReward/mean": 0.2305353358387947,
+      "rewards/CADChamferReward/std": 0.3032306879758835,
+      "rewards/CADFormatReward/mean": 0.9453125,
+      "rewards/CADFormatReward/std": 0.22850853204727173,
+      "rewards/CADInvalidReward/mean": 0.1171875,
+      "rewards/CADInvalidReward/std": 0.32395489513874054,
+      "rewards/CADProgressReward/mean": 0.004687500069849193,
+      "rewards/CADProgressReward/std": 0.030036812648177147,
+      "step": 2,
+      "step_time": 748.078124585445
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0546875,
+      "completions/max_length": 8543.5,
+      "completions/mean_length": 2519.375,
+      "completions/min_length": 698.5,
+      "epoch": 0.022222222222222223,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2393744403089944,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 2.1428571428571426e-07,
+      "loss": -0.08637851476669312,
+      "num_turns": 3.65625,
+      "reward": 1.0107279419898987,
+      "reward_std": 0.39696645736694336,
+      "rewards/CADCDValueReward/mean": 0.006078975275158882,
+      "rewards/CADCDValueReward/std": 0.015850992407649755,
+      "rewards/CADChamferReward/mean": 0.5208841860294342,
+      "rewards/CADChamferReward/std": 0.41364483535289764,
+      "rewards/CADFormatReward/mean": 0.9453125,
+      "rewards/CADFormatReward/std": 0.22850853204727173,
+      "rewards/CADInvalidReward/mean": 0.1484375,
+      "rewards/CADInvalidReward/std": 0.3567937761545181,
+      "rewards/CADProgressReward/mean": 0.034375001676380634,
+      "rewards/CADProgressReward/std": 0.07499999925494194,
+      "step": 3,
+      "step_time": 815.9138611884555
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 5940.5,
+      "completions/mean_length": 1743.0546875,
+      "completions/min_length": 569.5,
+      "epoch": 0.02962962962962963,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2627046329091631,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 2.857142857142857e-07,
+      "loss": -0.0766700804233551,
+      "num_turns": 3.296875,
+      "reward": 0.8810432255268097,
+      "reward_std": 0.3356814980506897,
+      "rewards/CADCDValueReward/mean": 0.008348886156454682,
+      "rewards/CADCDValueReward/std": 0.012812133878469467,
+      "rewards/CADChamferReward/mean": 0.36854322254657745,
+      "rewards/CADChamferReward/std": 0.3755342662334442,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0390625,
+      "rewards/CADInvalidReward/std": 0.13524486124515533,
+      "rewards/CADProgressReward/mean": 0.02500000037252903,
+      "rewards/CADProgressReward/std": 0.06597474217414856,
+      "step": 4,
+      "step_time": 744.047703415039
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6256.5,
+      "completions/mean_length": 1847.3828125,
+      "completions/min_length": 481.5,
+      "epoch": 0.037037037037037035,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.2720512949299147,
+      "gvpm/clamp_rate": 0.125,
+      "gvpm/mask_ratio": 0.125,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": -0.0666135624051094,
+      "num_turns": 3.3046875,
+      "reward": 0.896435558795929,
+      "reward_std": 0.29569074511528015,
+      "rewards/CADCDValueReward/mean": 0.01122749038040638,
+      "rewards/CADCDValueReward/std": 0.02446559350937605,
+      "rewards/CADChamferReward/mean": 0.3894043266773224,
+      "rewards/CADChamferReward/std": 0.4212627708911896,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.078125,
+      "rewards/CADInvalidReward/std": 0.27048972249031067,
+      "rewards/CADProgressReward/mean": 0.021875000558793545,
+      "rewards/CADProgressReward/std": 0.06271181628108025,
+      "step": 5,
+      "step_time": 705.8345783235272
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1825.5078125,
+      "completions/min_length": 478.5,
+      "epoch": 0.044444444444444446,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2598129047513917,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 4.285714285714285e-07,
+      "loss": -0.06650832295417786,
+      "num_turns": 3.1796875,
+      "reward": 0.9514619708061218,
+      "reward_std": 0.3243105411529541,
+      "rewards/CADCDValueReward/mean": 0.018256386276334524,
+      "rewards/CADCDValueReward/std": 0.03519853297621012,
+      "rewards/CADChamferReward/mean": 0.44208694994449615,
+      "rewards/CADChamferReward/std": 0.4390777051448822,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.04218750260770321,
+      "rewards/CADProgressReward/std": 0.08166900649666786,
+      "step": 6,
+      "step_time": 681.5786384145031
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 5864.0,
+      "completions/mean_length": 2106.703125,
+      "completions/min_length": 715.5,
+      "epoch": 0.05185185185185185,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.20583619133227324,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 5e-07,
+      "loss": -0.009767329320311546,
+      "num_turns": 3.484375,
+      "reward": 0.6793689131736755,
+      "reward_std": 0.28771600127220154,
+      "rewards/CADCDValueReward/mean": 0.019346370361745358,
+      "rewards/CADCDValueReward/std": 0.024582237005233765,
+      "rewards/CADChamferReward/mean": 0.1903063803911209,
+      "rewards/CADChamferReward/std": 0.31679461896419525,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.12198751419782639,
+      "rewards/CADInvalidReward/mean": 0.125,
+      "rewards/CADInvalidReward/std": 0.3333333432674408,
+      "rewards/CADProgressReward/mean": 0.009375000023283064,
+      "rewards/CADProgressReward/std": 0.039548974484205246,
+      "step": 7,
+      "step_time": 712.6712970329681
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 5095.5,
+      "completions/mean_length": 1699.3203125,
+      "completions/min_length": 465.5,
+      "epoch": 0.05925925925925926,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.26601978424045747,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 5.714285714285714e-07,
+      "loss": -0.030686555430293083,
+      "num_turns": 3.2109375,
+      "reward": 0.971459150314331,
+      "reward_std": 0.2936087101697922,
+      "rewards/CADCDValueReward/mean": 0.00972110778093338,
+      "rewards/CADCDValueReward/std": 0.014380504842847586,
+      "rewards/CADChamferReward/mean": 0.4503653943538666,
+      "rewards/CADChamferReward/std": 0.4412277638912201,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.1501840502023697,
+      "rewards/CADProgressReward/mean": 0.04218749888241291,
+      "rewards/CADProgressReward/std": 0.08035459369421005,
+      "step": 8,
+      "step_time": 700.7545191425015
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0390625,
+      "completions/max_length": 6546.0,
+      "completions/mean_length": 2595.953125,
+      "completions/min_length": 897.0,
+      "epoch": 0.06666666666666667,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.27073053298895466,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 6.428571428571429e-07,
+      "loss": -0.07535895705223083,
+      "num_turns": 3.6875,
+      "reward": 0.7963191270828247,
+      "reward_std": 0.33240535855293274,
+      "rewards/CADCDValueReward/mean": 0.009158202446997166,
+      "rewards/CADCDValueReward/std": 0.020484509877860546,
+      "rewards/CADChamferReward/mean": 0.3119441270828247,
+      "rewards/CADChamferReward/std": 0.3180331885814667,
+      "rewards/CADFormatReward/mean": 0.9609375,
+      "rewards/CADFormatReward/std": 0.13524486124515533,
+      "rewards/CADInvalidReward/mean": 0.1171875,
+      "rewards/CADInvalidReward/std": 0.3121146187186241,
+      "rewards/CADProgressReward/mean": 0.007812500232830644,
+      "rewards/CADProgressReward/std": 0.038841014727950096,
+      "step": 9,
+      "step_time": 813.0814969030325
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 2162.5234375,
+      "completions/min_length": 805.5,
+      "epoch": 0.07407407407407407,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.36699575014626085,
+      "gvpm/clamp_rate": 0.5625,
+      "gvpm/mask_ratio": 0.5625,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": -0.15485888719558716,
+      "num_turns": 3.4765625,
+      "reward": 0.8366019427776337,
+      "reward_std": 0.38951922953128815,
+      "rewards/CADCDValueReward/mean": 0.015387039165943861,
+      "rewards/CADCDValueReward/std": 0.026676760986447334,
+      "rewards/CADChamferReward/mean": 0.3311332166194916,
+      "rewards/CADChamferReward/std": 0.40434587001800537,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.1171875,
+      "rewards/CADInvalidReward/std": 0.32395489513874054,
+      "rewards/CADProgressReward/mean": 0.03437500074505806,
+      "rewards/CADProgressReward/std": 0.07593604549765587,
+      "step": 10,
+      "step_time": 735.5644311354845
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 4231.5,
+      "completions/mean_length": 1854.859375,
+      "completions/min_length": 562.0,
+      "epoch": 0.08148148148148149,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2706968612002178,
+      "gvpm/clamp_rate": 0.4375,
+      "gvpm/mask_ratio": 0.4375,
+      "learning_rate": 7.857142857142856e-07,
+      "loss": -0.05570477992296219,
+      "num_turns": 3.375,
+      "reward": 0.8794412016868591,
+      "reward_std": 0.27011872828006744,
+      "rewards/CADCDValueReward/mean": 0.005848402855917811,
+      "rewards/CADCDValueReward/std": 0.00792743917554617,
+      "rewards/CADChamferReward/mean": 0.37397244572639465,
+      "rewards/CADChamferReward/std": 0.3982246369123459,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.2925330847501755,
+      "rewards/CADProgressReward/mean": 0.010937500279396772,
+      "rewards/CADProgressReward/std": 0.045701704919338226,
+      "step": 11,
+      "step_time": 704.3677121474757
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 7880.0,
+      "completions/mean_length": 1946.6953125,
+      "completions/min_length": 541.0,
+      "epoch": 0.08888888888888889,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.27335119228951726,
+      "gvpm/clamp_rate": 0.125,
+      "gvpm/mask_ratio": 0.125,
+      "learning_rate": 8.57142857142857e-07,
+      "loss": -0.11574423313140869,
+      "num_turns": 3.25,
+      "reward": 0.9926949441432953,
+      "reward_std": 0.3828935921192169,
+      "rewards/CADCDValueReward/mean": 0.013384777121245861,
+      "rewards/CADCDValueReward/std": 0.023667567409574986,
+      "rewards/CADChamferReward/mean": 0.4809761345386505,
+      "rewards/CADChamferReward/std": 0.43272586166858673,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22292891144752502,
+      "rewards/CADProgressReward/mean": 0.0390625,
+      "rewards/CADProgressReward/std": 0.07929188758134842,
+      "step": 12,
+      "step_time": 641.6010130060604
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6145.0,
+      "completions/mean_length": 1847.7109375,
+      "completions/min_length": 546.5,
+      "epoch": 0.0962962962962963,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.321418259989506,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 9.285714285714285e-07,
+      "loss": -0.0536077544093132,
+      "num_turns": 3.3046875,
+      "reward": 0.8580175042152405,
+      "reward_std": 0.29149574041366577,
+      "rewards/CADCDValueReward/mean": 0.017112423665821552,
+      "rewards/CADCDValueReward/std": 0.027811897918581963,
+      "rewards/CADChamferReward/mean": 0.3533300310373306,
+      "rewards/CADChamferReward/std": 0.42514142394065857,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.1501840502023697,
+      "rewards/CADProgressReward/mean": 0.01718750037252903,
+      "rewards/CADProgressReward/std": 0.05642745830118656,
+      "step": 13,
+      "step_time": 648.2220064480789
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 5874.5,
+      "completions/mean_length": 1644.7265625,
+      "completions/min_length": 559.0,
+      "epoch": 0.1037037037037037,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.3468778785457278,
+      "gvpm/clamp_rate": 0.125,
+      "gvpm/mask_ratio": 0.125,
+      "learning_rate": 1e-06,
+      "loss": -0.11274628341197968,
+      "num_turns": 3.0859375,
+      "reward": 0.9221363961696625,
+      "reward_std": 0.3210814744234085,
+      "rewards/CADCDValueReward/mean": 0.009630282409489155,
+      "rewards/CADCDValueReward/std": 0.018014184664934874,
+      "rewards/CADChamferReward/mean": 0.4127614051103592,
+      "rewards/CADChamferReward/std": 0.39582359790802,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.0625,
+      "rewards/CADInvalidReward/std": 0.21978822350502014,
+      "rewards/CADProgressReward/mean": 0.02656250074505806,
+      "rewards/CADProgressReward/std": 0.06805390119552612,
+      "step": 14,
+      "step_time": 563.1225659624906
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 6446.0,
+      "completions/mean_length": 1554.78125,
+      "completions/min_length": 406.5,
+      "epoch": 0.1111111111111111,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.38267769429680915,
+      "gvpm/clamp_rate": 0.125,
+      "gvpm/mask_ratio": 0.125,
+      "learning_rate": 9.999623509195722e-07,
+      "loss": -0.10565665364265442,
+      "num_turns": 3.0859375,
+      "reward": 0.8347965478897095,
+      "reward_std": 0.3865255117416382,
+      "rewards/CADCDValueReward/mean": 0.02146594040095806,
+      "rewards/CADCDValueReward/std": 0.026481594890356064,
+      "rewards/CADChamferReward/mean": 0.3308902978897095,
+      "rewards/CADChamferReward/std": 0.4167514145374298,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.0703125,
+      "rewards/CADInvalidReward/std": 0.2572323754429817,
+      "rewards/CADProgressReward/mean": 0.023437500931322575,
+      "rewards/CADProgressReward/std": 0.06479097530245781,
+      "step": 15,
+      "step_time": 674.7883856399567
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1760.765625,
+      "completions/min_length": 592.5,
+      "epoch": 0.11851851851851852,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3103175454784033,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.99849409348102e-07,
+      "loss": -0.060198090970516205,
+      "num_turns": 3.2578125,
+      "reward": 0.8692543208599091,
+      "reward_std": 0.30393216013908386,
+      "rewards/CADCDValueReward/mean": 0.012020858936011791,
+      "rewards/CADCDValueReward/std": 0.020353389903903008,
+      "rewards/CADChamferReward/mean": 0.36925433576107025,
+      "rewards/CADChamferReward/std": 0.4049799144268036,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.109375,
+      "rewards/CADInvalidReward/std": 0.31043608486652374,
+      "rewards/CADProgressReward/mean": 0.023437500931322575,
+      "rewards/CADProgressReward/std": 0.06479097530245781,
+      "step": 16,
+      "step_time": 877.825865128485
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 4194.5,
+      "completions/mean_length": 1559.34375,
+      "completions/min_length": 447.5,
+      "epoch": 0.1259259259259259,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3371184986908411,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.996611922941747e-07,
+      "loss": -0.05865233764052391,
+      "num_turns": 3.1015625,
+      "reward": 1.0996561646461487,
+      "reward_std": 0.33785510063171387,
+      "rewards/CADCDValueReward/mean": 0.00933207105845213,
+      "rewards/CADCDValueReward/std": 0.02142926584929228,
+      "rewards/CADChamferReward/mean": 0.5777811706066132,
+      "rewards/CADChamferReward/std": 0.4316692054271698,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.1501840502023697,
+      "rewards/CADProgressReward/mean": 0.04375000111758709,
+      "rewards/CADProgressReward/std": 0.08254111930727959,
+      "step": 17,
+      "step_time": 645.3785339095048
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.03125,
+      "completions/max_length": 8694.5,
+      "completions/mean_length": 2302.1484375,
+      "completions/min_length": 492.5,
+      "epoch": 0.13333333333333333,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.28539059561280866,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.993977281025862e-07,
+      "loss": -0.076561838388443,
+      "num_turns": 3.484375,
+      "reward": 0.983812689781189,
+      "reward_std": 0.3841312676668167,
+      "rewards/CADCDValueReward/mean": 0.006533455103635788,
+      "rewards/CADCDValueReward/std": 0.014584028162062168,
+      "rewards/CADChamferReward/mean": 0.48537518084049225,
+      "rewards/CADChamferReward/std": 0.38363128900527954,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.17536810040473938,
+      "rewards/CADInvalidReward/mean": 0.109375,
+      "rewards/CADInvalidReward/std": 0.3145764470100403,
+      "rewards/CADProgressReward/mean": 0.02812500111758709,
+      "rewards/CADProgressReward/std": 0.07007648795843124,
+      "step": 18,
+      "step_time": 750.2256326780189
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8826.0,
+      "completions/mean_length": 1763.71875,
+      "completions/min_length": 552.0,
+      "epoch": 0.14074074074074075,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2941094542903187,
+      "gvpm/clamp_rate": 0.5,
+      "gvpm/mask_ratio": 0.5,
+      "learning_rate": 9.990590564500746e-07,
+      "loss": -0.056258171796798706,
+      "num_turns": 3.125,
+      "reward": 1.011709988117218,
+      "reward_std": 0.35970689356327057,
+      "rewards/CADCDValueReward/mean": 0.011950051411986351,
+      "rewards/CADCDValueReward/std": 0.02325180172920227,
+      "rewards/CADChamferReward/mean": 0.5046786963939667,
+      "rewards/CADChamferReward/std": 0.42988260090351105,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.0625,
+      "rewards/CADInvalidReward/std": 0.24176587909460068,
+      "rewards/CADProgressReward/mean": 0.03749999962747097,
+      "rewards/CADProgressReward/std": 0.07857595384120941,
+      "step": 19,
+      "step_time": 689.71919516247
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1588.125,
+      "completions/min_length": 488.5,
+      "epoch": 0.14814814814814814,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2945812214682008,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.986452283393451e-07,
+      "loss": -0.09122254699468613,
+      "num_turns": 3.0390625,
+      "reward": 0.8520500063896179,
+      "reward_std": 0.3548283874988556,
+      "rewards/CADCDValueReward/mean": 0.01876841951161623,
+      "rewards/CADCDValueReward/std": 0.030105404555797577,
+      "rewards/CADChamferReward/mean": 0.3512687683105469,
+      "rewards/CADChamferReward/std": 0.40571722388267517,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.046875,
+      "rewards/CADInvalidReward/std": 0.20967156440019608,
+      "rewards/CADProgressReward/mean": 0.01718750037252903,
+      "rewards/CADProgressReward/std": 0.05642745643854141,
+      "step": 20,
+      "step_time": 668.8609365549055
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 5942.0,
+      "completions/mean_length": 1437.421875,
+      "completions/min_length": 425.5,
+      "epoch": 0.15555555555555556,
+      "frac_reward_zero_std": 0.1875,
+      "grad_norm": 0.2422842709739306,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 9.981563060913889e-07,
+      "loss": -0.03578227385878563,
+      "num_turns": 2.859375,
+      "reward": 1.0359516143798828,
+      "reward_std": 0.25851575285196304,
+      "rewards/CADCDValueReward/mean": 0.010054457932710648,
+      "rewards/CADCDValueReward/std": 0.020625630393624306,
+      "rewards/CADChamferReward/mean": 0.5242328643798828,
+      "rewards/CADChamferReward/std": 0.46253058314323425,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.015625,
+      "rewards/CADInvalidReward/std": 0.08768405020236969,
+      "rewards/CADProgressReward/mean": 0.031249999068677425,
+      "rewards/CADProgressReward/std": 0.0720081739127636,
+      "step": 21,
+      "step_time": 565.7626083085197
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 2412.03125,
+      "completions/min_length": 924.0,
+      "epoch": 0.16296296296296298,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2306572381778335,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.975923633360984e-07,
+      "loss": -0.07813145220279694,
+      "num_turns": 3.6171875,
+      "reward": 0.8644551336765289,
+      "reward_std": 0.3148357570171356,
+      "rewards/CADCDValueReward/mean": 0.0105775217525661,
+      "rewards/CADCDValueReward/std": 0.019866307266056538,
+      "rewards/CADChamferReward/mean": 0.3644551634788513,
+      "rewards/CADChamferReward/std": 0.37886635959148407,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.0859375,
+      "rewards/CADInvalidReward/std": 0.2792757377028465,
+      "rewards/CADProgressReward/mean": 0.023437500931322575,
+      "rewards/CADProgressReward/std": 0.06441672891378403,
+      "step": 22,
+      "step_time": 894.0683379225084
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 4819.5,
+      "completions/mean_length": 1781.421875,
+      "completions/min_length": 390.0,
+      "epoch": 0.17037037037037037,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2569209241985606,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 9.96953485001178e-07,
+      "loss": -0.13267382979393005,
+      "num_turns": 3.3828125,
+      "reward": 1.056757390499115,
+      "reward_std": 0.30872842669487,
+      "rewards/CADCDValueReward/mean": 0.008921411354094744,
+      "rewards/CADCDValueReward/std": 0.018665974959731102,
+      "rewards/CADChamferReward/mean": 0.5434761941432953,
+      "rewards/CADChamferReward/std": 0.4205426275730133,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.02656250074505806,
+      "rewards/CADProgressReward/std": 0.06837157532572746,
+      "step": 23,
+      "step_time": 838.0536323420238
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6497.0,
+      "completions/mean_length": 2079.6484375,
+      "completions/min_length": 720.0,
+      "epoch": 0.17777777777777778,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.25247172099517995,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.96239767299355e-07,
+      "loss": -0.07227814197540283,
+      "num_turns": 3.640625,
+      "reward": 1.0185624957084656,
+      "reward_std": 0.3580661565065384,
+      "rewards/CADCDValueReward/mean": 0.006263577379286289,
+      "rewards/CADCDValueReward/std": 0.01340128155425191,
+      "rewards/CADChamferReward/mean": 0.5115312337875366,
+      "rewards/CADChamferReward/std": 0.3675832748413086,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.078125,
+      "rewards/CADInvalidReward/std": 0.2543507218360901,
+      "rewards/CADProgressReward/mean": 0.029687500558793545,
+      "rewards/CADProgressReward/std": 0.07079743221402168,
+      "step": 24,
+      "step_time": 723.5349406970199
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6039.0,
+      "completions/mean_length": 1637.6484375,
+      "completions/min_length": 402.5,
+      "epoch": 0.18518518518518517,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.33133232062197165,
+      "gvpm/clamp_rate": 0.5625,
+      "gvpm/mask_ratio": 0.5625,
+      "learning_rate": 9.9545131771389e-07,
+      "loss": -0.11317317932844162,
+      "num_turns": 3.1875,
+      "reward": 1.029203176498413,
+      "reward_std": 0.3615288585424423,
+      "rewards/CADCDValueReward/mean": 0.01225523091852665,
+      "rewards/CADCDValueReward/std": 0.02145096193999052,
+      "rewards/CADChamferReward/mean": 0.515140637755394,
+      "rewards/CADChamferReward/std": 0.4296441823244095,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22292891144752502,
+      "rewards/CADProgressReward/mean": 0.03593750111758709,
+      "rewards/CADProgressReward/std": 0.07714678719639778,
+      "step": 25,
+      "step_time": 616.4760901704431
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.03125,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1895.828125,
+      "completions/min_length": 482.0,
+      "epoch": 0.1925925925925926,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3176523032612676,
+      "gvpm/clamp_rate": 0.75,
+      "gvpm/mask_ratio": 0.75,
+      "learning_rate": 9.945882549823904e-07,
+      "loss": -0.05554371699690819,
+      "num_turns": 3.3515625,
+      "reward": 0.8108876049518585,
+      "reward_std": 0.3796968460083008,
+      "rewards/CADCDValueReward/mean": 0.01634395821020007,
+      "rewards/CADCDValueReward/std": 0.027299873530864716,
+      "rewards/CADChamferReward/mean": 0.3187001198530197,
+      "rewards/CADChamferReward/std": 0.4155423492193222,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.16902101784944534,
+      "rewards/CADInvalidReward/mean": 0.1171875,
+      "rewards/CADInvalidReward/std": 0.32395489513874054,
+      "rewards/CADProgressReward/mean": 0.015625,
+      "rewards/CADProgressReward/std": 0.054097944870591164,
+      "step": 26,
+      "step_time": 784.9924379324657
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.046875,
+      "completions/max_length": 8604.0,
+      "completions/mean_length": 2056.6640625,
+      "completions/min_length": 621.0,
+      "epoch": 0.2,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.264799714555237,
+      "gvpm/clamp_rate": 0.0,
+      "gvpm/mask_ratio": 0.0,
+      "learning_rate": 9.936507090789292e-07,
+      "loss": -0.061725594103336334,
+      "num_turns": 3.28125,
+      "reward": 1.0490643382072449,
+      "reward_std": 0.3362731784582138,
+      "rewards/CADCDValueReward/mean": 0.006162935635074973,
+      "rewards/CADCDValueReward/std": 0.014397912658751011,
+      "rewards/CADChamferReward/mean": 0.5553143322467804,
+      "rewards/CADChamferReward/std": 0.38265109062194824,
+      "rewards/CADFormatReward/mean": 0.953125,
+      "rewards/CADFormatReward/std": 0.20967156440019608,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.2925330847501755,
+      "rewards/CADProgressReward/mean": 0.03437500074505806,
+      "rewards/CADProgressReward/std": 0.07605084776878357,
+      "step": 27,
+      "step_time": 719.3276593934861
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 2122.7421875,
+      "completions/min_length": 783.5,
+      "epoch": 0.2074074074074074,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.31592059791818744,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.926388211944704e-07,
+      "loss": -0.08546764403581619,
+      "num_turns": 3.53125,
+      "reward": 0.9773533344268799,
+      "reward_std": 0.38023094832897186,
+      "rewards/CADCDValueReward/mean": 0.012124743778258562,
+      "rewards/CADCDValueReward/std": 0.025397044606506824,
+      "rewards/CADChamferReward/mean": 0.47813452780246735,
+      "rewards/CADChamferReward/std": 0.4035791605710983,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.109375,
+      "rewards/CADInvalidReward/std": 0.31043608486652374,
+      "rewards/CADProgressReward/mean": 0.021875000093132257,
+      "rewards/CADProgressReward/std": 0.06099376082420349,
+      "step": 28,
+      "step_time": 725.8918347665458
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 2083.890625,
+      "completions/min_length": 564.0,
+      "epoch": 0.21481481481481482,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.30025494404400316,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.915527437156081e-07,
+      "loss": -0.0366211012005806,
+      "num_turns": 3.5078125,
+      "reward": 0.9832450449466705,
+      "reward_std": 0.35620221495628357,
+      "rewards/CADCDValueReward/mean": 0.0032473006285727024,
+      "rewards/CADCDValueReward/std": 0.004252071725204587,
+      "rewards/CADChamferReward/mean": 0.47933876514434814,
+      "rewards/CADChamferReward/std": 0.37349840998649597,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.2925330847501755,
+      "rewards/CADProgressReward/mean": 0.023437500931322575,
+      "rewards/CADProgressReward/std": 0.06479097902774811,
+      "step": 29,
+      "step_time": 787.6940262585413
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 4964.5,
+      "completions/mean_length": 1841.96875,
+      "completions/min_length": 529.5,
+      "epoch": 0.2222222222222222,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.27354791426998043,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.90392640201615e-07,
+      "loss": -0.059005916118621826,
+      "num_turns": 3.3984375,
+      "reward": 0.9467656910419464,
+      "reward_std": 0.2836601436138153,
+      "rewards/CADCDValueReward/mean": 0.010562224313616753,
+      "rewards/CADCDValueReward/std": 0.02116239257156849,
+      "rewards/CADChamferReward/mean": 0.43817196786403656,
+      "rewards/CADChamferReward/std": 0.390131875872612,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.015625,
+      "rewards/CADInvalidReward/std": 0.08768405020236969,
+      "rewards/CADProgressReward/mean": 0.017187499906867743,
+      "rewards/CADProgressReward/std": 0.055855147540569305,
+      "step": 30,
+      "step_time": 644.6225378446397
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 6864.5,
+      "completions/mean_length": 1962.7890625,
+      "completions/min_length": 524.5,
+      "epoch": 0.22962962962962963,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.258167449583024,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.891586853598138e-07,
+      "loss": -0.07502081990242004,
+      "num_turns": 3.1328125,
+      "reward": 0.9736192226409912,
+      "reward_std": 0.31906336545944214,
+      "rewards/CADCDValueReward/mean": 0.005501617211848497,
+      "rewards/CADCDValueReward/std": 0.012062105350196362,
+      "rewards/CADChamferReward/mean": 0.47986921668052673,
+      "rewards/CADChamferReward/std": 0.37294983863830566,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.10652101784944534,
+      "rewards/CADInvalidReward/mean": 0.0625,
+      "rewards/CADInvalidReward/std": 0.24176587909460068,
+      "rewards/CADProgressReward/mean": 0.010937500046566129,
+      "rewards/CADProgressReward/std": 0.044585783034563065,
+      "step": 31,
+      "step_time": 652.3193422019831
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 5661.0,
+      "completions/mean_length": 1726.6953125,
+      "completions/min_length": 604.5,
+      "epoch": 0.23703703703703705,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.2729273570380658,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.878510650192642e-07,
+      "loss": -0.05360779911279678,
+      "num_turns": 3.1328125,
+      "reward": 1.009830892086029,
+      "reward_std": 0.27592554688453674,
+      "rewards/CADCDValueReward/mean": 0.010024767369031906,
+      "rewards/CADCDValueReward/std": 0.022714214399456978,
+      "rewards/CADChamferReward/mean": 0.5020184218883514,
+      "rewards/CADChamferReward/std": 0.4194524884223938,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.10652101784944534,
+      "rewards/CADProgressReward/mean": 0.031250000931322575,
+      "rewards/CADProgressReward/std": 0.0730636678636074,
+      "step": 32,
+      "step_time": 610.3340677325032
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6038.5,
+      "completions/mean_length": 1980.5625,
+      "completions/min_length": 351.0,
+      "epoch": 0.24444444444444444,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.3199160524824189,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.8646997610278e-07,
+      "loss": -0.06718609482049942,
+      "num_turns": 3.375,
+      "reward": 0.8588527143001556,
+      "reward_std": 0.29866546392440796,
+      "rewards/CADCDValueReward/mean": 0.005571319488808513,
+      "rewards/CADCDValueReward/std": 0.008845038712024689,
+      "rewards/CADChamferReward/mean": 0.3557277321815491,
+      "rewards/CADChamferReward/std": 0.344767764210701,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.20939241349697113,
+      "rewards/CADProgressReward/mean": 0.014062500093132257,
+      "rewards/CADProgressReward/std": 0.051446475088596344,
+      "step": 33,
+      "step_time": 721.2837582180509
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6548.5,
+      "completions/mean_length": 1799.359375,
+      "completions/min_length": 566.5,
+      "epoch": 0.2518518518518518,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.28745768475077366,
+      "gvpm/clamp_rate": 0.5625,
+      "gvpm/mask_ratio": 0.5625,
+      "learning_rate": 9.85015626597272e-07,
+      "loss": -0.0849800705909729,
+      "num_turns": 3.3046875,
+      "reward": 0.8399341702461243,
+      "reward_std": 0.28047215938568115,
+      "rewards/CADCDValueReward/mean": 0.012512448243796825,
+      "rewards/CADCDValueReward/std": 0.020015444606542587,
+      "rewards/CADChamferReward/mean": 0.33212170004844666,
+      "rewards/CADChamferReward/std": 0.41149912774562836,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.0625,
+      "rewards/CADInvalidReward/std": 0.24176587909460068,
+      "rewards/CADProgressReward/mean": 0.0234375,
+      "rewards/CADProgressReward/std": 0.06441672891378403,
+      "step": 34,
+      "step_time": 743.4361940279487
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1665.359375,
+      "completions/min_length": 444.5,
+      "epoch": 0.25925925925925924,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.31312486042551846,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.83488235522426e-07,
+      "loss": -0.0926455706357956,
+      "num_turns": 3.015625,
+      "reward": 1.0441208481788635,
+      "reward_std": 0.43992879986763,
+      "rewards/CADCDValueReward/mean": 0.010189782828092575,
+      "rewards/CADCDValueReward/std": 0.020550732035189867,
+      "rewards/CADChamferReward/mean": 0.5284958332777023,
+      "rewards/CADChamferReward/std": 0.42446738481521606,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.05468750186264515,
+      "rewards/CADProgressReward/std": 0.08842188864946365,
+      "step": 35,
+      "step_time": 662.2956334860064
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 7357.5,
+      "completions/mean_length": 1978.1484375,
+      "completions/min_length": 509.5,
+      "epoch": 0.26666666666666666,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.29081684621406195,
+      "gvpm/clamp_rate": 0.125,
+      "gvpm/mask_ratio": 0.125,
+      "learning_rate": 9.8188803289772e-07,
+      "loss": -0.08506050705909729,
+      "num_turns": 3.265625,
+      "reward": 1.035053789615631,
+      "reward_std": 0.295712873339653,
+      "rewards/CADCDValueReward/mean": 0.006636728765442967,
+      "rewards/CADCDValueReward/std": 0.017200497444719076,
+      "rewards/CADChamferReward/mean": 0.5288038104772568,
+      "rewards/CADChamferReward/std": 0.438412606716156,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.10652101784944534,
+      "rewards/CADInvalidReward/mean": 0.0859375,
+      "rewards/CADInvalidReward/std": 0.2628752738237381,
+      "rewards/CADProgressReward/mean": 0.03593749925494194,
+      "rewards/CADProgressReward/std": 0.07714678719639778,
+      "step": 36,
+      "step_time": 808.6509420100483
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.03125,
+      "completions/max_length": 8671.5,
+      "completions/mean_length": 2072.84375,
+      "completions/min_length": 559.5,
+      "epoch": 0.2740740740740741,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3353417284638203,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.802152597077828e-07,
+      "loss": -0.07194460928440094,
+      "num_turns": 3.3671875,
+      "reward": 1.0173083543777466,
+      "reward_std": 0.40730439126491547,
+      "rewards/CADCDValueReward/mean": 0.004399771336466074,
+      "rewards/CADCDValueReward/std": 0.008308042772114277,
+      "rewards/CADChamferReward/mean": 0.5134020745754242,
+      "rewards/CADChamferReward/std": 0.3815064877271652,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.17536810040473938,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.29378482699394226,
+      "rewards/CADProgressReward/mean": 0.03906250186264515,
+      "rewards/CADProgressReward/std": 0.07969209179282188,
+      "step": 37,
+      "step_time": 753.216583872505
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 3918.0,
+      "completions/mean_length": 1757.5703125,
+      "completions/min_length": 453.5,
+      "epoch": 0.2814814814814815,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.2589390600255359,
+      "gvpm/clamp_rate": 0.4375,
+      "gvpm/mask_ratio": 0.4375,
+      "learning_rate": 9.784701678661044e-07,
+      "loss": -0.05084498971700668,
+      "num_turns": 3.2890625,
+      "reward": 0.8897316157817841,
+      "reward_std": 0.24061861634254456,
+      "rewards/CADCDValueReward/mean": 0.01003743102774024,
+      "rewards/CADCDValueReward/std": 0.01767158042639494,
+      "rewards/CADChamferReward/mean": 0.3787941187620163,
+      "rewards/CADChamferReward/std": 0.3944964110851288,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.16902101784944534,
+      "rewards/CADProgressReward/mean": 0.021875000558793545,
+      "rewards/CADProgressReward/std": 0.06271182000637054,
+      "step": 38,
+      "step_time": 626.9985265335417
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 5941.5,
+      "completions/mean_length": 1885.8515625,
+      "completions/min_length": 486.0,
+      "epoch": 0.28888888888888886,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.33316093398613267,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 9.766530201770967e-07,
+      "loss": -0.10036209970712662,
+      "num_turns": 3.4609375,
+      "reward": 0.9943822026252747,
+      "reward_std": 0.38469935953617096,
+      "rewards/CADCDValueReward/mean": 0.010812665335834026,
+      "rewards/CADCDValueReward/std": 0.023920376785099506,
+      "rewards/CADChamferReward/mean": 0.48422594368457794,
+      "rewards/CADChamferReward/std": 0.4070392996072769,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.0859375,
+      "rewards/CADInvalidReward/std": 0.28213727474212646,
+      "rewards/CADProgressReward/mean": 0.03593750111758709,
+      "rewards/CADProgressReward/std": 0.07670491188764572,
+      "step": 39,
+      "step_time": 718.7788579615299
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1818.8515625,
+      "completions/min_length": 481.0,
+      "epoch": 0.2962962962962963,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.31176581838680095,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.747640902965182e-07,
+      "loss": -0.04899011552333832,
+      "num_turns": 3.109375,
+      "reward": 1.0272006392478943,
+      "reward_std": 0.39842282235622406,
+      "rewards/CADCDValueReward/mean": 0.006507785525172949,
+      "rewards/CADCDValueReward/std": 0.017087630927562714,
+      "rewards/CADChamferReward/mean": 0.5279819071292877,
+      "rewards/CADChamferReward/std": 0.4242442846298218,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.2925330847501755,
+      "rewards/CADProgressReward/mean": 0.02187499962747097,
+      "rewards/CADProgressReward/std": 0.06208721734583378,
+      "step": 40,
+      "step_time": 768.1774763509748
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 5956.0,
+      "completions/mean_length": 1333.609375,
+      "completions/min_length": 453.0,
+      "epoch": 0.3037037037037037,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.24401950758696947,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.728036626902607e-07,
+      "loss": -0.03415040299296379,
+      "num_turns": 2.6875,
+      "reward": 1.1090074181556702,
+      "reward_std": 0.2634388953447342,
+      "rewards/CADCDValueReward/mean": 0.005830263951793313,
+      "rewards/CADCDValueReward/std": 0.009733580984175205,
+      "rewards/CADChamferReward/mean": 0.6004136800765991,
+      "rewards/CADChamferReward/std": 0.4362885504961014,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.10652101784944534,
+      "rewards/CADProgressReward/mean": 0.03281250037252903,
+      "rewards/CADProgressReward/std": 0.07437803223729134,
+      "step": 41,
+      "step_time": 581.5830892570084
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1627.2265625,
+      "completions/min_length": 508.5,
+      "epoch": 0.3111111111111111,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.2133396915352835,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.707720325915103e-07,
+      "loss": -0.028079399839043617,
+      "num_turns": 2.921875,
+      "reward": 0.9877127707004547,
+      "reward_std": 0.24002548307180405,
+      "rewards/CADCDValueReward/mean": 0.00834473967552185,
+      "rewards/CADCDValueReward/std": 0.015562626533210278,
+      "rewards/CADChamferReward/mean": 0.48458781838417053,
+      "rewards/CADChamferReward/std": 0.42006009817123413,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.16902101784944534,
+      "rewards/CADProgressReward/mean": 0.021875000558793545,
+      "rewards/CADProgressReward/std": 0.06271181628108025,
+      "step": 42,
+      "step_time": 611.8008063075249
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8680.0,
+      "completions/mean_length": 1774.984375,
+      "completions/min_length": 567.0,
+      "epoch": 0.31851851851851853,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.3286262792890542,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.686695059562874e-07,
+      "loss": -0.06646941602230072,
+      "num_turns": 3.265625,
+      "reward": 0.9156647622585297,
+      "reward_std": 0.3191141039133072,
+      "rewards/CADCDValueReward/mean": 0.009640823118388653,
+      "rewards/CADCDValueReward/std": 0.016673089005053043,
+      "rewards/CADChamferReward/mean": 0.40550853312015533,
+      "rewards/CADChamferReward/std": 0.42234233021736145,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.17536810040473938,
+      "rewards/CADProgressReward/mean": 0.03593750111758709,
+      "rewards/CADProgressReward/std": 0.07714678719639778,
+      "step": 43,
+      "step_time": 663.2222139424994
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.046875,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 2027.703125,
+      "completions/min_length": 562.0,
+      "epoch": 0.32592592592592595,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.3098968483848029,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.664963994173693e-07,
+      "loss": -0.05407789349555969,
+      "num_turns": 3.1171875,
+      "reward": 0.907115250825882,
+      "reward_std": 0.3342723697423935,
+      "rewards/CADCDValueReward/mean": 0.009987911209464073,
+      "rewards/CADCDValueReward/std": 0.015833669807761908,
+      "rewards/CADChamferReward/mean": 0.41649025678634644,
+      "rewards/CADChamferReward/std": 0.3898242264986038,
+      "rewards/CADFormatReward/mean": 0.953125,
+      "rewards/CADFormatReward/std": 0.21304203569889069,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.2925330847501755,
+      "rewards/CADProgressReward/mean": 0.02812500111758709,
+      "rewards/CADProgressReward/std": 0.06948306784033775,
+      "step": 44,
+      "step_time": 686.5821760474937
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 5765.5,
+      "completions/mean_length": 1786.7109375,
+      "completions/min_length": 556.5,
+      "epoch": 0.3333333333333333,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.29315669725592414,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.642530402366078e-07,
+      "loss": 0.0072112735360860825,
+      "num_turns": 3.3125,
+      "reward": 0.990590512752533,
+      "reward_std": 0.32899370789527893,
+      "rewards/CADCDValueReward/mean": 0.010082908440381289,
+      "rewards/CADCDValueReward/std": 0.01896606106311083,
+      "rewards/CADChamferReward/mean": 0.4819967597723007,
+      "rewards/CADChamferReward/std": 0.4468446969985962,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.09375,
+      "rewards/CADInvalidReward/std": 0.2925330847501755,
+      "rewards/CADProgressReward/mean": 0.03281250037252903,
+      "rewards/CADProgressReward/std": 0.07462168112397194,
+      "step": 45,
+      "step_time": 670.6647546880995
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 9148.0,
+      "completions/mean_length": 2010.7734375,
+      "completions/min_length": 527.5,
+      "epoch": 0.34074074074074073,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.26375500508995686,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.619397662556433e-07,
+      "loss": -0.0605773963034153,
+      "num_turns": 3.3203125,
+      "reward": 1.0370360612869263,
+      "reward_std": 0.34883715212345123,
+      "rewards/CADCDValueReward/mean": 0.005773921264335513,
+      "rewards/CADCDValueReward/std": 0.01515254145488143,
+      "rewards/CADChamferReward/mean": 0.5339110493659973,
+      "rewards/CADChamferReward/std": 0.3714523911476135,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.02968750149011612,
+      "rewards/CADProgressReward/std": 0.06992901489138603,
+      "step": 46,
+      "step_time": 781.8355874035042
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 6850.5,
+      "completions/mean_length": 2329.1640625,
+      "completions/min_length": 578.5,
+      "epoch": 0.34814814814814815,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.25034575161954825,
+      "gvpm/clamp_rate": 0.4375,
+      "gvpm/mask_ratio": 0.4375,
+      "learning_rate": 9.595569258450289e-07,
+      "loss": -0.03825327008962631,
+      "num_turns": 3.515625,
+      "reward": 0.7986271977424622,
+      "reward_std": 0.287222221493721,
+      "rewards/CADCDValueReward/mean": 0.011657494585961103,
+      "rewards/CADCDValueReward/std": 0.02588548604398966,
+      "rewards/CADChamferReward/mean": 0.30878348648548126,
+      "rewards/CADChamferReward/std": 0.3143278509378433,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.10652101784944534,
+      "rewards/CADInvalidReward/mean": 0.109375,
+      "rewards/CADInvalidReward/std": 0.3145764470100403,
+      "rewards/CADProgressReward/mean": 0.0031250000465661287,
+      "rewards/CADProgressReward/std": 0.02500000223517418,
+      "step": 47,
+      "step_time": 685.2994031485287
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 6124.5,
+      "completions/mean_length": 1663.7734375,
+      "completions/min_length": 458.0,
+      "epoch": 0.35555555555555557,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.28837179688759407,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.571048778517652e-07,
+      "loss": -0.0845719426870346,
+      "num_turns": 3.140625,
+      "reward": 1.0179510712623596,
+      "reward_std": 0.28746291995048523,
+      "rewards/CADCDValueReward/mean": 0.010810192907229066,
+      "rewards/CADCDValueReward/std": 0.023981335572898388,
+      "rewards/CADChamferReward/mean": 0.5140447914600372,
+      "rewards/CADChamferReward/std": 0.38752706348896027,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.08768405020236969,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.1501840502023697,
+      "rewards/CADProgressReward/mean": 0.0234375,
+      "rewards/CADProgressReward/std": 0.06364523060619831,
+      "step": 48,
+      "step_time": 639.4493279859889
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6029.5,
+      "completions/mean_length": 1708.6953125,
+      "completions/min_length": 528.0,
+      "epoch": 0.362962962962963,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.2624397697673845,
+      "gvpm/clamp_rate": 0.375,
+      "gvpm/mask_ratio": 0.375,
+      "learning_rate": 9.545839915452611e-07,
+      "loss": 0.0031496845185756683,
+      "num_turns": 3.078125,
+      "reward": 0.8401894569396973,
+      "reward_std": 0.2741198167204857,
+      "rewards/CADCDValueReward/mean": 0.00806389469653368,
+      "rewards/CADCDValueReward/std": 0.011322741396725178,
+      "rewards/CADChamferReward/mean": 0.3370644748210907,
+      "rewards/CADChamferReward/std": 0.39350321888923645,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.0703125,
+      "rewards/CADInvalidReward/std": 0.24497227370738983,
+      "rewards/CADProgressReward/mean": 0.021875000558793545,
+      "rewards/CADProgressReward/std": 0.06208721548318863,
+      "step": 49,
+      "step_time": 654.7832974030753
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 4623.0,
+      "completions/mean_length": 1640.65625,
+      "completions/min_length": 515.5,
+      "epoch": 0.37037037037037035,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.33510029382490514,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.519946465617217e-07,
+      "loss": -0.0020774416625499725,
+      "num_turns": 3.09375,
+      "reward": 1.146395742893219,
+      "reward_std": 0.36798277497291565,
+      "rewards/CADCDValueReward/mean": 0.005812318995594978,
+      "rewards/CADCDValueReward/std": 0.017129408195614815,
+      "rewards/CADChamferReward/mean": 0.6221769750118256,
+      "rewards/CADChamferReward/std": 0.3998716175556183,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.04843750037252903,
+      "rewards/CADProgressReward/std": 0.0861823633313179,
+      "step": 50,
+      "step_time": 639.4267683760263
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.03125,
+      "completions/max_length": 7139.5,
+      "completions/mean_length": 1918.0390625,
+      "completions/min_length": 487.0,
+      "epoch": 0.37777777777777777,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3791754505354697,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.493372328469769e-07,
+      "loss": -0.10320444405078888,
+      "num_turns": 3.1796875,
+      "reward": 0.8491643965244293,
+      "reward_std": 0.3237500786781311,
+      "rewards/CADCDValueReward/mean": 0.007575992261990905,
+      "rewards/CADCDValueReward/std": 0.01029264833778143,
+      "rewards/CADChamferReward/mean": 0.3530706316232681,
+      "rewards/CADChamferReward/std": 0.3840659260749817,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.12198751419782639,
+      "rewards/CADInvalidReward/mean": 0.0625,
+      "rewards/CADInvalidReward/std": 0.24176587909460068,
+      "rewards/CADProgressReward/mean": 0.02343750186264515,
+      "rewards/CADProgressReward/std": 0.06479097902774811,
+      "step": 51,
+      "step_time": 592.0421478134813
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8761.0,
+      "completions/mean_length": 1802.5625,
+      "completions/min_length": 428.5,
+      "epoch": 0.3851851851851852,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.31925761072226416,
+      "gvpm/clamp_rate": 0.5,
+      "gvpm/mask_ratio": 0.5,
+      "learning_rate": 9.466121505977576e-07,
+      "loss": -0.0613069087266922,
+      "num_turns": 3.1015625,
+      "reward": 0.9655038118362427,
+      "reward_std": 0.340903177857399,
+      "rewards/CADCDValueReward/mean": 0.007181169930845499,
+      "rewards/CADCDValueReward/std": 0.012167679611593485,
+      "rewards/CADChamferReward/mean": 0.45925378799438477,
+      "rewards/CADChamferReward/std": 0.43096397817134857,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.0703125,
+      "rewards/CADInvalidReward/std": 0.2534134313464165,
+      "rewards/CADProgressReward/mean": 0.02812500111758709,
+      "rewards/CADProgressReward/std": 0.06871827319264412,
+      "step": 52,
+      "step_time": 752.4245209319633
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6691.0,
+      "completions/mean_length": 1669.578125,
+      "completions/min_length": 589.0,
+      "epoch": 0.3925925925925926,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2601323689782601,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.43819810201427e-07,
+      "loss": -0.06911112368106842,
+      "num_turns": 2.921875,
+      "reward": 1.141762375831604,
+      "reward_std": 0.28559741377830505,
+      "rewards/CADCDValueReward/mean": 0.007342529483139515,
+      "rewards/CADCDValueReward/std": 0.01919363532215357,
+      "rewards/CADChamferReward/mean": 0.6245749294757843,
+      "rewards/CADChamferReward/std": 0.3975260406732559,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.015625,
+      "rewards/CADInvalidReward/std": 0.08768405020236969,
+      "rewards/CADProgressReward/mean": 0.04218750074505806,
+      "rewards/CADProgressReward/std": 0.08166900277137756,
+      "step": 53,
+      "step_time": 562.2636092515313
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1722.53125,
+      "completions/min_length": 392.5,
+      "epoch": 0.4,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3375879590417734,
+      "gvpm/clamp_rate": 0.0,
+      "gvpm/mask_ratio": 0.0,
+      "learning_rate": 9.409606321741774e-07,
+      "loss": -0.013662035576999187,
+      "num_turns": 3.03125,
+      "reward": 1.099386751651764,
+      "reward_std": 0.2912728488445282,
+      "rewards/CADCDValueReward/mean": 0.006153033580631018,
+      "rewards/CADCDValueReward/std": 0.015778191853314638,
+      "rewards/CADChamferReward/mean": 0.592355489730835,
+      "rewards/CADChamferReward/std": 0.4214763343334198,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.16902101784944534,
+      "rewards/CADProgressReward/mean": 0.029687500558793545,
+      "rewards/CADProgressReward/std": 0.07135875523090363,
+      "step": 54,
+      "step_time": 650.6152248754515
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.03125,
+      "completions/max_length": 8643.5,
+      "completions/mean_length": 2059.3203125,
+      "completions/min_length": 632.5,
+      "epoch": 0.4074074074074074,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3226498289188323,
+      "gvpm/clamp_rate": 0.4375,
+      "gvpm/mask_ratio": 0.4375,
+      "learning_rate": 9.380350470977032e-07,
+      "loss": -0.07155250757932663,
+      "num_turns": 3.5,
+      "reward": 0.9070218205451965,
+      "reward_std": 0.3886219561100006,
+      "rewards/CADCDValueReward/mean": 0.011722303926944733,
+      "rewards/CADCDValueReward/std": 0.025446307845413685,
+      "rewards/CADChamferReward/mean": 0.41092802584171295,
+      "rewards/CADChamferReward/std": 0.38270294666290283,
+      "rewards/CADFormatReward/mean": 0.96875,
+      "rewards/CADFormatReward/std": 0.16902101784944534,
+      "rewards/CADInvalidReward/mean": 0.140625,
+      "rewards/CADInvalidReward/std": 0.3496479392051697,
+      "rewards/CADProgressReward/mean": 0.023437500931322575,
+      "rewards/CADProgressReward/std": 0.06479097902774811,
+      "step": 55,
+      "step_time": 770.8545176874613
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 7938.0,
+      "completions/mean_length": 2245.65625,
+      "completions/min_length": 591.5,
+      "epoch": 0.4148148148148148,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3003170703202294,
+      "gvpm/clamp_rate": 0.75,
+      "gvpm/mask_ratio": 0.75,
+      "learning_rate": 9.350434955543557e-07,
+      "loss": -0.06011253967881203,
+      "num_turns": 3.5859375,
+      "reward": 0.8453920781612396,
+      "reward_std": 0.3145531862974167,
+      "rewards/CADCDValueReward/mean": 0.01054031541571021,
+      "rewards/CADCDValueReward/std": 0.02487563155591488,
+      "rewards/CADChamferReward/mean": 0.33757956326007843,
+      "rewards/CADChamferReward/std": 0.33816616237163544,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.125,
+      "rewards/CADInvalidReward/std": 0.3333333432674408,
+      "rewards/CADProgressReward/mean": 0.0234375,
+      "rewards/CADProgressReward/std": 0.06364523060619831,
+      "step": 56,
+      "step_time": 909.908546074992
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 7579.5,
+      "completions/mean_length": 1823.109375,
+      "completions/min_length": 408.0,
+      "epoch": 0.4222222222222222,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.3200746381649447,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.319864280607934e-07,
+      "loss": -0.05908238887786865,
+      "num_turns": 3.25,
+      "reward": 1.0055773258209229,
+      "reward_std": 0.3037809878587723,
+      "rewards/CADCDValueReward/mean": 0.011222207453101873,
+      "rewards/CADCDValueReward/std": 0.024534993804991245,
+      "rewards/CADChamferReward/mean": 0.5055772811174393,
+      "rewards/CADChamferReward/std": 0.3964270204305649,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.046875,
+      "rewards/CADInvalidReward/std": 0.20967156440019608,
+      "rewards/CADProgressReward/mean": 0.015625,
+      "rewards/CADProgressReward/std": 0.05409794673323631,
+      "step": 57,
+      "step_time": 756.2379550259211
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8471.5,
+      "completions/mean_length": 1748.890625,
+      "completions/min_length": 585.5,
+      "epoch": 0.42962962962962964,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3429335487801308,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.28864305000136e-07,
+      "loss": -0.0015911739319562912,
+      "num_turns": 3.0390625,
+      "reward": 1.049176573753357,
+      "reward_std": 0.3722991943359375,
+      "rewards/CADCDValueReward/mean": 0.00502880965359509,
+      "rewards/CADCDValueReward/std": 0.00911300745792687,
+      "rewards/CADChamferReward/mean": 0.5390203297138214,
+      "rewards/CADChamferReward/std": 0.3947935700416565,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.0234375,
+      "rewards/CADInvalidReward/std": 0.1501840502023697,
+      "rewards/CADProgressReward/mean": 0.03593750111758709,
+      "rewards/CADProgressReward/std": 0.07714678719639778,
+      "step": 58,
+      "step_time": 714.4565632169833
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6798.0,
+      "completions/mean_length": 1776.546875,
+      "completions/min_length": 473.5,
+      "epoch": 0.43703703703703706,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.2650050271247873,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 9.256775965526326e-07,
+      "loss": -0.050642773509025574,
+      "num_turns": 3.3046875,
+      "reward": 0.9662282764911652,
+      "reward_std": 0.24029946327209473,
+      "rewards/CADCDValueReward/mean": 0.007690779631957412,
+      "rewards/CADCDValueReward/std": 0.016633108258247375,
+      "rewards/CADChamferReward/mean": 0.45841577649116516,
+      "rewards/CADChamferReward/std": 0.42017658054828644,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.046875,
+      "rewards/CADInvalidReward/std": 0.21304203569889069,
+      "rewards/CADProgressReward/mean": 0.023437500931322575,
+      "rewards/CADProgressReward/std": 0.06479097530245781,
+      "step": 59,
+      "step_time": 703.0263737409841
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 5316.5,
+      "completions/mean_length": 1924.828125,
+      "completions/min_length": 593.5,
+      "epoch": 0.4444444444444444,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2728441543145627,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.224267826248536e-07,
+      "loss": -0.0760185718536377,
+      "num_turns": 3.3046875,
+      "reward": 0.9995194375514984,
+      "reward_std": 0.36176206171512604,
+      "rewards/CADCDValueReward/mean": 0.008715164847671986,
+      "rewards/CADCDValueReward/std": 0.022952506318688393,
+      "rewards/CADChamferReward/mean": 0.4815506935119629,
+      "rewards/CADChamferReward/std": 0.40026241540908813,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.16902101784944534,
+      "rewards/CADProgressReward/mean": 0.03593750111758709,
+      "rewards/CADProgressReward/std": 0.07714678719639778,
+      "step": 60,
+      "step_time": 644.8002563769696
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8472.0,
+      "completions/mean_length": 2038.0390625,
+      "completions/min_length": 575.0,
+      "epoch": 0.45185185185185184,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.21574088610134093,
+      "gvpm/clamp_rate": 0.4375,
+      "gvpm/mask_ratio": 0.4375,
+      "learning_rate": 9.191123527774189e-07,
+      "loss": -0.06047581881284714,
+      "num_turns": 3.3828125,
+      "reward": 0.9609400033950806,
+      "reward_std": 0.3012053519487381,
+      "rewards/CADCDValueReward/mean": 0.007246244233101606,
+      "rewards/CADCDValueReward/std": 0.015529151540249586,
+      "rewards/CADChamferReward/mean": 0.45859625935554504,
+      "rewards/CADChamferReward/std": 0.414530873298645,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.0859375,
+      "rewards/CADInvalidReward/std": 0.2792757377028465,
+      "rewards/CADProgressReward/mean": 0.028125000186264515,
+      "rewards/CADProgressReward/std": 0.069929588586092,
+      "step": 61,
+      "step_time": 705.3993218655232
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 5964.0,
+      "completions/mean_length": 1686.34375,
+      "completions/min_length": 568.5,
+      "epoch": 0.45925925925925926,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.28915098737436373,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.157348061512726e-07,
+      "loss": -0.024678653106093407,
+      "num_turns": 3.1796875,
+      "reward": 0.964135080575943,
+      "reward_std": 0.319868266582489,
+      "rewards/CADCDValueReward/mean": 0.005077847046777606,
+      "rewards/CADCDValueReward/std": 0.008623983478173614,
+      "rewards/CADChamferReward/mean": 0.45944756269454956,
+      "rewards/CADChamferReward/std": 0.3756243586540222,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22292891144752502,
+      "rewards/CADProgressReward/mean": 0.01718750037252903,
+      "rewards/CADProgressReward/std": 0.05642745830118656,
+      "step": 62,
+      "step_time": 673.9347062259912
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 3676.0,
+      "completions/mean_length": 1735.6171875,
+      "completions/min_length": 523.0,
+      "epoch": 0.4666666666666667,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3370947637179292,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.122946513925126e-07,
+      "loss": -0.058004140853881836,
+      "num_turns": 3.3203125,
+      "reward": 0.9642046689987183,
+      "reward_std": 0.37482963502407074,
+      "rewards/CADCDValueReward/mean": 0.015951286535710096,
+      "rewards/CADCDValueReward/std": 0.037708403542637825,
+      "rewards/CADChamferReward/mean": 0.4509234130382538,
+      "rewards/CADChamferReward/std": 0.3834534287452698,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.1015625,
+      "rewards/CADInvalidReward/std": 0.30191153287887573,
+      "rewards/CADProgressReward/mean": 0.026562499813735485,
+      "rewards/CADProgressReward/std": 0.06837157532572746,
+      "step": 63,
+      "step_time": 662.9919982755673
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 3434.0,
+      "completions/mean_length": 1800.2421875,
+      "completions/min_length": 700.0,
+      "epoch": 0.4740740740740741,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.34419399180551935,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 9.087924065757918e-07,
+      "loss": -0.02677270956337452,
+      "num_turns": 3.25,
+      "reward": 1.0374482870101929,
+      "reward_std": 0.334428071975708,
+      "rewards/CADCDValueReward/mean": 0.00554040796123445,
+      "rewards/CADCDValueReward/std": 0.01146679138764739,
+      "rewards/CADChamferReward/mean": 0.5218233019113541,
+      "rewards/CADChamferReward/std": 0.34231896698474884,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0078125,
+      "rewards/CADInvalidReward/std": 0.0625,
+      "rewards/CADProgressReward/mean": 0.03125,
+      "rewards/CADProgressReward/std": 0.06974460370838642,
+      "step": 64,
+      "step_time": 589.4235910624848
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 6190.0,
+      "completions/mean_length": 2119.671875,
+      "completions/min_length": 610.5,
+      "epoch": 0.48148148148148145,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.2546129693778692,
+      "gvpm/clamp_rate": 0.25,
+      "gvpm/mask_ratio": 0.25,
+      "learning_rate": 9.052285991262974e-07,
+      "loss": -0.03790409862995148,
+      "num_turns": 3.5234375,
+      "reward": 0.9496057033538818,
+      "reward_std": 0.28532589972019196,
+      "rewards/CADCDValueReward/mean": 0.004761902382597327,
+      "rewards/CADCDValueReward/std": 0.009021487087011337,
+      "rewards/CADChamferReward/mean": 0.4394494444131851,
+      "rewards/CADChamferReward/std": 0.39276182651519775,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.078125,
+      "rewards/CADInvalidReward/std": 0.27048972249031067,
+      "rewards/CADProgressReward/mean": 0.02812500111758709,
+      "rewards/CADProgressReward/std": 0.06948306784033775,
+      "step": 65,
+      "step_time": 666.4829357709968
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0078125,
+      "completions/max_length": 5897.5,
+      "completions/mean_length": 1852.3515625,
+      "completions/min_length": 582.0,
+      "epoch": 0.4888888888888889,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.32647079367496457,
+      "gvpm/clamp_rate": 0.3125,
+      "gvpm/mask_ratio": 0.3125,
+      "learning_rate": 9.016037657403223e-07,
+      "loss": -0.07062443345785141,
+      "num_turns": 3.3671875,
+      "reward": 1.0007648468017578,
+      "reward_std": 0.320096418261528,
+      "rewards/CADCDValueReward/mean": 0.003650993457995355,
+      "rewards/CADCDValueReward/std": 0.004244861658662558,
+      "rewards/CADChamferReward/mean": 0.4859210401773453,
+      "rewards/CADChamferReward/std": 0.3747712969779968,
+      "rewards/CADFormatReward/mean": 0.9921875,
+      "rewards/CADFormatReward/std": 0.0625,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.16902101784944534,
+      "rewards/CADProgressReward/mean": 0.037500000558793545,
+      "rewards/CADProgressReward/std": 0.07773387432098389,
+      "step": 66,
+      "step_time": 747.0061523441109
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 3447.5,
+      "completions/mean_length": 1537.765625,
+      "completions/min_length": 559.5,
+      "epoch": 0.4962962962962963,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3684932883824493,
+      "gvpm/clamp_rate": 0.4375,
+      "gvpm/mask_ratio": 0.4375,
+      "learning_rate": 8.979184523044418e-07,
+      "loss": -0.08982193470001221,
+      "num_turns": 3.203125,
+      "reward": 1.005136787891388,
+      "reward_std": 0.4044913351535797,
+      "rewards/CADCDValueReward/mean": 0.015445917844772339,
+      "rewards/CADCDValueReward/std": 0.03328438941389322,
+      "rewards/CADChamferReward/mean": 0.4887305647134781,
+      "rewards/CADChamferReward/std": 0.41610924899578094,
+      "rewards/CADFormatReward/mean": 1.0,
+      "rewards/CADFormatReward/std": 0.0,
+      "rewards/CADInvalidReward/mean": 0.0625,
+      "rewards/CADInvalidReward/std": 0.24176587909460068,
+      "rewards/CADProgressReward/mean": 0.03281250037252903,
+      "rewards/CADProgressReward/std": 0.07462168112397194,
+      "step": 67,
+      "step_time": 612.9340318380273
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 5851.0,
+      "completions/mean_length": 1642.3359375,
+      "completions/min_length": 412.0,
+      "epoch": 0.5037037037037037,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.32089011997272,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 8.941732138133031e-07,
+      "loss": -0.06648235768079758,
+      "num_turns": 2.953125,
+      "reward": 1.0746761560440063,
+      "reward_std": 0.3606071174144745,
+      "rewards/CADCDValueReward/mean": 0.00935483630746603,
+      "rewards/CADCDValueReward/std": 0.022447858937084675,
+      "rewards/CADChamferReward/mean": 0.570769876241684,
+      "rewards/CADChamferReward/std": 0.41204652190208435,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.10652101784944534,
+      "rewards/CADInvalidReward/mean": 0.03125,
+      "rewards/CADInvalidReward/std": 0.16902101784944534,
+      "rewards/CADProgressReward/mean": 0.031250000931322575,
+      "rewards/CADProgressReward/std": 0.07267312332987785,
+      "step": 68,
+      "step_time": 626.2576917275437
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.0234375,
+      "completions/max_length": 8577.0,
+      "completions/mean_length": 1886.4140625,
+      "completions/min_length": 558.5,
+      "epoch": 0.5111111111111111,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.3337271280587593,
+      "gvpm/clamp_rate": 0.0625,
+      "gvpm/mask_ratio": 0.0625,
+      "learning_rate": 8.903686142860471e-07,
+      "loss": -0.07734221965074539,
+      "num_turns": 3.484375,
+      "reward": 0.8702357113361359,
+      "reward_std": 0.33238084614276886,
+      "rewards/CADCDValueReward/mean": 0.010502251796424389,
+      "rewards/CADCDValueReward/std": 0.023212259635329247,
+      "rewards/CADChamferReward/mean": 0.3733607083559036,
+      "rewards/CADChamferReward/std": 0.40685878694057465,
+      "rewards/CADFormatReward/mean": 0.9765625,
+      "rewards/CADFormatReward/std": 0.1501840502023697,
+      "rewards/CADInvalidReward/mean": 0.1484375,
+      "rewards/CADInvalidReward/std": 0.3567937761545181,
+      "rewards/CADProgressReward/mean": 0.01718750037252903,
+      "rewards/CADProgressReward/std": 0.05642745830118656,
+      "step": 69,
+      "step_time": 859.1491852009203
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.015625,
+      "completions/max_length": 8192.0,
+      "completions/mean_length": 1653.1640625,
+      "completions/min_length": 455.0,
+      "epoch": 0.5185185185185185,
+      "frac_reward_zero_std": 0.0625,
+      "grad_norm": 0.3461346197894174,
+      "gvpm/clamp_rate": 0.1875,
+      "gvpm/mask_ratio": 0.1875,
+      "learning_rate": 8.865052266813685e-07,
+      "loss": -0.10868415236473083,
+      "num_turns": 3.21875,
+      "reward": 1.0690314769744873,
+      "reward_std": 0.35482974350452423,
+      "rewards/CADCDValueReward/mean": 0.00583080283831805,
+      "rewards/CADCDValueReward/std": 0.012201638892292976,
+      "rewards/CADChamferReward/mean": 0.5573127865791321,
+      "rewards/CADChamferReward/std": 0.41786521673202515,
+      "rewards/CADFormatReward/mean": 0.984375,
+      "rewards/CADFormatReward/std": 0.125,
+      "rewards/CADInvalidReward/mean": 0.0546875,
+      "rewards/CADInvalidReward/std": 0.22850853204727173,
+      "rewards/CADProgressReward/mean": 0.0390625,
+      "rewards/CADProgressReward/std": 0.07989032194018364,
+      "step": 70,
+      "step_time": 692.2213933565654
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 270,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd18413c4ff5e7c850bca903ece116a208a810e6b886b84c0c6535f8cad55167
+size 13137

zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info("Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info("Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)