Merge pull request #5290 from simonJJJ/qwen2_vl

support qwen2-vl Former-commit-id: 7156f832af8505b26371559d340c0e69eb962bbc
2024-08-30 02:10:36 +08:00
parent c1369a1ec9 7466fd4387
commit 77c2c7076b
8 changed files with 185 additions and 6 deletions
--- a/src/llamafactory/model/model_utils/misc.py
+++ b/src/llamafactory/model/model_utils/misc.py
@@ -36,6 +36,8 @@ def find_all_linear_modules(model: "PreTrainedModel", freeze_vision_tower: bool)
        forbidden_modules.add("output")
    elif model.config.model_type in ["llava", "paligemma"]:
        forbidden_modules.add("multi_modal_projector")
+    elif model.config.model_type in ["qwen2_vl"]:
+        forbidden_modules.add("merger")

    if freeze_vision_tower:
        forbidden_modules.add("vision_tower")