423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-10-14 15:52:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

[data] fix qwen2audio plugin (#7166)

Browse Source 
* Update pairwise.py

[data]Repair multimodal model dpo training

* Update pairwise.py

[data]repair multimodal model dpo training using deepcopy

* Update pairwise.py

* Update mm_plugin.py

Former-commit-id: 86763dfdb8e9e5668c1ddd7e924e4be76bf78368
This commit is contained in:
sirui.li 2025-03-05 18:03:36 +08:00 committed by GitHub
parent 31f9daa362
commit fd0357a26d
1 changed files with 1 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
src/llamafactory/data/mm_plugin.py
View File

@ -993,6 +993,7 @@ class Qwen2AudioPlugin(BasePlugin):
self._validate_input(processor, images, videos, audios)
bos_token: str = getattr(processor, "audio_bos_token")
eos_token: str = getattr(processor, "audio_eos_token")
messages = deepcopy(messages)
mm_inputs = self._get_mm_inputs([], [], audios, processor)
if "feature_attention_mask" in mm_inputs:
audio_lengths = mm_inputs["feature_attention_mask"].sum(-1).tolist()