File size: 1,929 Bytes
1f57542 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 | export HTTP_PROXY=http://sys-proxy-rd-relay.byted.org:8118
export http_proxy=http://sys-proxy-rd-relay.byted.org:8118
export https_proxy=http://sys-proxy-rd-relay.byted.org:8118
export no_proxy="$no_proxy,.byteintl.net"
export HF_ENDPOINT=https://hf-mirror.com
cd quantization_metric/
model_id=/mnt/bn/life-mllm/users/cxr/quantization/models/Qwen/Qwen2.5-14B
model_name=$(basename "$model_id")
cuda_id=4
model=/mnt/bn/life-mllm/users/cxr/quantization/models/${model_name}-quantization-layer-mlp
# output_dir=Alpha_values_mlp
# bit_layers_dir=/mnt/bn/life-mllm/users/cxr/quantization/quantization_metric/bit_layers
# result_dir=/mnt/bn/life-mllm/users/cxr/quantization/lm-evaluation-harness/results
start=$(date +%s.%N)
# rm -rf $model
modes=("mlp")
for mode in ${modes[@]}; do
for idx in {32..47}; do
echo $mode $idx
cd ../quantization_metric
python -u main.py --bit_layer_idx $idx --save_dir ${model} --mode $mode --model_id $model_id --cuda_id $cuda_id
cd ../lm-evaluation-harness
bash run_scripts/eval_base_qwen2_5_14b.sh ${model} ${mode}_${idx}
rm -rf ${model}
end=$(date +%s.%N)
runtime=$(awk "BEGIN {print $end - $start}")
echo "Execution time: $runtime seconds"
done
done
model=/mnt/bn/life-mllm/users/cxr/quantization/models/${model_name}-quantization-layer
start=$(date +%s.%N)
# rm -rf $model
modes=("self_attn")
for mode in ${modes[@]}; do
for idx in {32..47}; do
echo $mode $idx
cd ../quantization_metric
python -u main.py --bit_layer_idx $idx --save_dir ${model} --mode $mode --model_id $model_id --cuda_id $cuda_id
cd ../lm-evaluation-harness
bash run_scripts/eval_base_qwen2_5_14b.sh ${model} ${mode}_${idx}
rm -rf ${model}
end=$(date +%s.%N)
runtime=$(awk "BEGIN {print $end - $start}")
echo "Execution time: $runtime seconds"
done
done |