kyy/llm_trainer
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add forbidden modules

Browse Source
This commit is contained in:
hiyouga committed 2024-11-23 18:34:15 +00:00
1 parent 446441fdb0
commit df477370dc
2 files changed
+6 -2
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

src/llamafactory/model/model_utils/misc.py
+4 -2
View File
@@ -34,13 +34,15 @@ def find_all_linear_modules(model: "PreTrainedModel", freeze_vision_tower: bool)
forbidden_modules.add("output_layer")
elif model_type == "internlm2":
forbidden_modules.add("output")
elif model_type in ["llava", "llava_next", "llava_next_video", "paligemma", "video_llava"]:
elif model_type in ["llava", "llava_next", "llava_next_video", "mllama", "paligemma", "video_llava"]:
forbidden_modules.add("multi_modal_projector")
elif model_type == "qwen2_vl":
forbidden_modules.add("merger")
if freeze_vision_tower:
if model_type == "qwen2_vl":
if model_type == "mllama":
forbidden_modules.add("vision_model")
elif model_type == "qwen2_vl":
forbidden_modules.add("visual")
else:
forbidden_modules.add("vision_tower")
src/llamafactory/model/model_utils/visual.py
+2
View File
@@ -191,6 +191,8 @@ def patch_target_modules(
if finetuning_args.freeze_vision_tower:
if model_type in ["llava", "llava_next", "llava_next_video", "paligemma", "pixtral", "video_llava"]:
return "^(?!.*vision_tower).*(?:{}).*".format("|".join(target_modules))
elif model_type == "mllama":
return "^(?!.*vision_model).*(?:{}).*".format("|".join(target_modules))
elif model_type == "qwen2_vl":
return "^(?!.*visual).*(?:{}).*".format("|".join(target_modules))
else: