if past_key_value is not None and q_len == 1:

This commit is contained in:
Daniel Han-Chen 2024-01-31 03:36:40 +11:00
commit 55fe6052ca
2 changed files with 2 additions and 2 deletions

View file

@ -310,7 +310,7 @@ def LlamaAttention_fast_forward(
bsz, q_len, _ = hidden_states.size()
# Check for inference
if past_key_value is not None and q_len == 1 and bsz == 1:
if past_key_value is not None and q_len == 1:
A, past_key_value = LlamaAttention_fast_forward_inference(
self,
hidden_states,

View file

@ -49,7 +49,7 @@ def MistralAttention_fast_forward(
bsz, q_len, _ = hidden_states.size()
# Check for inference
if True:
if past_key_value is not None and q_len == 1:
A, past_key_value = LlamaAttention_fast_forward_inference(
self,
hidden_states,