423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2026-03-14 07:55:59 +08:00
Code Issues Packages Projects Releases Wiki Activity

[fix] fallback to audio_processor when feature_extractor is missing (#10267)

Browse Source 
Co-authored-by: kevin <742971636@qq.com>
This commit is contained in:
xxddccaa
2026-03-10 19:36:41 +08:00
committed by GitHub
parent a3d44e3152
commit 3045adf0ba
1 changed files with 9 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
src/llamafactory/data/mm_plugin.py
View File

@@ -161,7 +161,9 @@ class MMPluginMixin:
video_processor: BaseImageProcessor = getattr( video_processor: BaseImageProcessor = getattr(
processor, "video_processor", getattr(processor, "image_processor", None) processor, "video_processor", getattr(processor, "image_processor", None)
) )
feature_extractor: SequenceFeatureExtractor = getattr(processor, "feature_extractor", None) feature_extractor: SequenceFeatureExtractor = getattr(processor, "feature_extractor", None) or getattr(
processor, "audio_processor", None
)
if len(images) != 0 and self.image_token is None: if len(images) != 0 and self.image_token is None:
raise ValueError( raise ValueError(
"This model does not support image input. Please check whether the correct `template` is used." "This model does not support image input. Please check whether the correct `template` is used."
@@ -390,7 +392,9 @@ class MMPluginMixin:
mm_inputs.update(video_processor(videos, return_tensors="pt")) mm_inputs.update(video_processor(videos, return_tensors="pt"))
if len(audios) != 0: if len(audios) != 0:
feature_extractor: SequenceFeatureExtractor = getattr(processor, "feature_extractor", None) feature_extractor: SequenceFeatureExtractor = getattr(processor, "feature_extractor", None) or getattr(
processor, "audio_processor", None
)
audios = self._regularize_audios( audios = self._regularize_audios(
audios, audios,
sampling_rate=getattr(processor, "audio_sampling_rate", 16000), sampling_rate=getattr(processor, "audio_sampling_rate", 16000),
@@ -1876,7 +1880,9 @@ class Qwen2OmniPlugin(Qwen2VLPlugin):
) -> dict[str, "torch.Tensor"]: ) -> dict[str, "torch.Tensor"]:
image_processor: BaseImageProcessor = getattr(processor, "image_processor", None) image_processor: BaseImageProcessor = getattr(processor, "image_processor", None)
video_processor: BaseVideoProcessor = getattr(processor, "video_processor", None) video_processor: BaseVideoProcessor = getattr(processor, "video_processor", None)
feature_extractor: SequenceFeatureExtractor = getattr(processor, "feature_extractor", None) feature_extractor: SequenceFeatureExtractor = getattr(processor, "feature_extractor", None) or getattr(
processor, "audio_processor", None
)
mm_inputs = {} mm_inputs = {}
if len(images) != 0: if len(images) != 0:
images = self._regularize_images( images = self._regularize_images(