# AutoAWQ 0.2.5 two-device error break points
# b /home/justin/work/AutoAWQ/awq/models/base.py:170
# b /home/justin/work/AutoAWQ/awq/quantize/quantizer.py:61

# debug GPU memory measuring issue
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/eval/bench.py:242

# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/adapter/autoawq.py:39
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/adapter/autoawq.py:46
# b /home/justin/miniconda3/envs/quant-eval/lib/python3.11/site-packages/transformers/modeling_utils.py:955, param_device == 'cpu'
# b /home/justin/miniconda3/envs/quant-eval/lib/python3.11/site-packages/transformers/modeling_utils.py:973, param_device == 'cpu'
# b /home/justin/work/transformers/modeling_utils.py:989
# b /home/justin/work/transformers/src/transformers/modeling_utils.py:4795
# b /home/justin/work/transformers/src/transformers/modeling_utils.py:4213
# b /home/justin/work/AutoAWQ/awq/models/base.py:338
# b /home/justin/work/accelerate/src/accelerate/big_modeling.py:420
# b /home/justin/work/accelerate/src/accelerate/hooks.py:579, module_name == 'model.layers.28.mlp.gate_proj'
# b /home/justin/work/accelerate/src/accelerate/hooks.py:579, module_name == 'model.layers.28'
# b /home/justin/work/accelerate/src/accelerate/hooks.py:288
# b /home/justin/work/accelerate/src/accelerate/hooks.py:308
# b /home/justin/work/accelerate/src/accelerate/hooks.py:327
# b /home/justin/miniconda3/envs/quant-eval/lib/python3.11/site-packages/transformers/modeling_utils.py:4194
# b /home/justin/miniconda3/envs/quant-eval/lib/python3.11/site-packages/transformers/quantizers/quantizer_bnb_4bit.py:190
# b /home/justin/work/transformers/src/transformers/modeling_utils.py:4194
# b /home/justin/work/hqq/hqq/utils/optimizer.py:21

# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/misc/qweight.py:216
# b /home/justin/work/lm-quant-toolkit/src/cli.py:436
# b /home/justin/work/hqq/hqq/utils/optimizer.py:19
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/prep/sensitivity.py:206, i == 25
# b /home/justin/work/hqq/hqq/models/base.py:262
# b /home/justin/work/hqq/hqq/utils/optimizer.py:26
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/prep/sensitivity.py:395
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/prep/sensitivity.py:413
# b /home/justin/work/hqq/hqq/utils/optimizer.py:26
# b /home/justin/work/hqq/hqq/core/quantize.py:116
# b /home/justin/work/hqq/hqq/core/quantize.py:175
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/prep/sensitivity.py:24
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/eval/bench.py:256
# b /home/justin/work/lm-quant-toolkit/src/lm_quant_toolkit/eval/bench.py:270
# b /home/justin/work/hqq/hqq/utils/optimizer.py:157
b /home/justin/work/CLIP_benchmark/clip_benchmark/models/open_clip_hqq.py:13
b /home/justin/work/hqq/hqq/models/open_clip/base.py:63