423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-17 04:10:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix up

Browse Source
This commit is contained in:
hiyouga
2024-07-15 01:04:56 +08:00
parent 15b399a82f
commit 29ebcd75d5
19 changed files with 47 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/llamafactory/model/model_utils/longlora.py
View File

@@ -71,8 +71,6 @@ def llama_attention_forward(
cos, sin = self.rotary_emb(value_states, position_ids)
query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
past_key_value = getattr(self, "past_key_value", past_key_value)
if past_key_value is not None:
cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
@@ -156,8 +154,6 @@ def llama_flash_attention_2_forward(
cos, sin = self.rotary_emb(value_states, position_ids)
query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
past_key_value = getattr(self, "past_key_value", past_key_value)
if past_key_value is not None:
cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)