423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-18 12:50:38 +08:00
Code Issues Packages Projects Releases Wiki Activity

[model] add deepstack_merger_list to Qwen3-VL vision_model_keys (#9399)

Browse Source
This commit is contained in:
한송민
2025-11-04 20:27:34 +09:00
committed by GitHub
parent 934b3084ee
commit 5a9939050e
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/llamafactory/model/model_utils/visual.py
View File

@@ -355,7 +355,7 @@ _register_composite_model(
_register_composite_model(
model_type="qwen3_vl",
projector_key="visual.merger",
vision_model_keys=["visual.patch_embed", "visual.blocks"],
vision_model_keys=["visual.patch_embed", "visual.blocks", "visual.deepstack_merger_list"],
language_model_keys=["language_model", "lm_head"],
lora_conflict_keys=["patch_embed"],
)
@@ -364,7 +364,7 @@ _register_composite_model(
_register_composite_model(
model_type="qwen3_vl_moe",
projector_key="visual.merger",
vision_model_keys=["visual.patch_embed", "visual.blocks"],
vision_model_keys=["visual.patch_embed", "visual.blocks", "visual.deepstack_merger_list"],
language_model_keys=["language_model", "lm_head"],
lora_conflict_keys=["patch_embed"],
)
@@ -373,7 +373,7 @@ _register_composite_model(
_register_composite_model(
model_type="qwen3_omni_moe_thinker",
projector_key="visual.merger",
vision_model_keys=["visual.patch_embed", "visual.blocks", "audio_tower"],
vision_model_keys=["visual.patch_embed", "visual.blocks", "visual.deepstack_merger_list", "audio_tower"],
language_model_keys=["model", "lm_head"],
lora_conflict_keys=["patch_embed"],
)