Yang Wang 8f1c9488b5 handle missing scale_inv_name (#2) ... * handle missing scale_inv_name Fixed an issue where `weight` and `weight_scale_inv` (e.g. `model.layers.39.mlp.experts.92.gate_proj.weight` and `model.layers.39.mlp.experts.92.gate_proj.weight_scale_inv`) were not in the same SafeTensor, causing an assertion error due to scale_inv_name not being in the state_dict. * sort filename to reduce memory costs * Add CUDA cache clearing in memory management Added torch.cuda.empty_cache() to free up unused memory on the GPU,		2024-12-27 09:34:38 +08:00
..
configs	Release DeepSeek-V3	2024-12-26 19:01:57 +08:00
convert.py	Release DeepSeek-V3	2024-12-26 19:01:57 +08:00
fp8_cast_bf16.py	handle missing scale_inv_name (#2)	2024-12-27 09:34:38 +08:00
generate.py	Release DeepSeek-V3	2024-12-26 19:01:57 +08:00
kernel.py	Release DeepSeek-V3	2024-12-26 19:01:57 +08:00
model.py	Release DeepSeek-V3	2024-12-26 19:01:57 +08:00
requirements.txt	Release DeepSeek-V3	2024-12-26 19:01:57 +08:00