ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update llama.py

Browse source
This commit is contained in:
Daniel Han-Chen 2024-03-10 13:11:42 +11:00
commit baf8e4c0a8
1 changed files with 2 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

5
unsloth/models/llama.py
View file

@ -705,8 +705,7 @@ def CausalLM_fast_forward(fast_forward_inference):
*args, **kwargs,
) -> Union[Tuple, CausalLMOutputWithPast]:
if False: #past_key_values is not None and \
hasattr(self.model.layers[0].self_attn, "paged_attention"):
if False: #past_key_values is not None and hasattr(self.model.layers[0].self_attn, "paged_attention"):
outputs = fast_forward_inference(
self.model,
input_ids,
@ -723,7 +722,7 @@ def CausalLM_fast_forward(fast_forward_inference):
# decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
self.model._has_no_labels = labels is None
outputs = self.model(
input_ids=input_ids,
causal_mask=causal_mask,