keyError: 'sdpa'

#3
by fengzi258 - opened

When loading the model: DeepSeek-V2, i got the following error:

File "/root/miniconda3/envs/eval-python3.9/lib/python3.9/site-packages/transformers/modeling_utils.py", line 3550, in from_pretrained
model = cls(config, *model_args, **model_kwargs)
File "/root/.cache/huggingface/modules/transformers_modules/DeepSeek-V2/modeling_deepseek.py", line 1588, in init
self.model = DeepseekV2Model(config)
File "/root/.cache/huggingface/modules/transformers_modules/DeepSeek-V2/modeling_deepseek.py", line 1404, in init
[
File "/root/.cache/huggingface/modules/transformers_modules/DeepSeek-V2/modeling_deepseek.py", line 1405, in
DeepseekV2DecoderLayer(config, layer_idx)
File "/root/.cache/huggingface/modules/transformers_modules/DeepSeek-V2/modeling_deepseek.py", line 1187, in init
self.self_attn = ATTENTION_CLASSES[config._attn_implementation](
KeyError: 'sdpa'

Sign up or log in to comment