ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix Llama model save for full param train

Browse Source 
Former-commit-id: ca17b5db4f97c3ec9fe2004877f150e8f51ab4b5
This commit is contained in:
marko1616
2024-03-30 23:45:04 +08:00
parent 6198121923
commit a6858a36c0
2 changed files with 9 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/llmtuner/model/patcher.py
View File

@@ -312,6 +312,15 @@ def patch_config(
def patch_model(
model: "PreTrainedModel", tokenizer: "PreTrainedTokenizer", model_args: "ModelArguments", is_trainable: bool
) -> None:
#Config check and fix
gen_config = model.generation_config
if not gen_config.do_sample and (
(gen_config.temperature is not None and gen_config.temperature != 1.0)
or (gen_config.top_p is not None and gen_config.top_p != 1.0)
or (gen_config.typical_p is not None and gen_config.typical_p != 1.0)
):
gen_config.do_sample = True
if "GenerationMixin" not in str(model.generate.__func__):
model.generate = MethodType(PreTrainedModel.generate, model)