423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-02 11:42:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

tiny fix

Browse Source 
Former-commit-id: 27776c34741ca0c58ed793bcdf1acd5e4a81fb39
This commit is contained in:
hiyouga 2024-03-31 00:10:29 +08:00
parent 91067408f5
commit 3cf35e57db
1 changed files with 8 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
src/llmtuner/model/patcher.py
View File

@ -312,15 +312,6 @@ def patch_config(
def patch_model( def patch_model(
model: "PreTrainedModel", tokenizer: "PreTrainedTokenizer", model_args: "ModelArguments", is_trainable: bool model: "PreTrainedModel", tokenizer: "PreTrainedTokenizer", model_args: "ModelArguments", is_trainable: bool
) -> None: ) -> 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__): if "GenerationMixin" not in str(model.generate.__func__):
model.generate = MethodType(PreTrainedModel.generate, model) model.generate = MethodType(PreTrainedModel.generate, model)
@ -328,6 +319,14 @@ def patch_model(
setattr(model, "lm_head", model.transformer.output_layer) setattr(model, "lm_head", model.transformer.output_layer)
setattr(model, "_keys_to_ignore_on_save", ["lm_head.weight"]) setattr(model, "_keys_to_ignore_on_save", ["lm_head.weight"])
gen_config = model.generation_config # check and fix 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 model_args.resize_vocab: if model_args.resize_vocab:
_resize_embedding_layer(model, tokenizer) _resize_embedding_layer(model, tokenizer)