423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-29 18:20:35 +08:00
Code Issues Packages Projects Releases Wiki Activity

[model] unsloth resume from checkpoint bug (#8423)

Browse Source 
Co-authored-by: viyer <vivek_iyer2@apple.com>
This commit is contained in:
Vivek Iyer
2025-06-23 04:43:54 -04:00
committed by GitHub
parent 1a67692886
commit 4e0efdb48b
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/llamafactory/model/model_utils/unsloth.py
View File

@@ -80,12 +80,12 @@ def get_unsloth_peft_model(
def load_unsloth_peft_model(
config: "PretrainedConfig", model_args: "ModelArguments", is_trainable: bool
config: "PretrainedConfig", model_args: "ModelArguments", finetuning_args: "FinetuningArguments", is_trainable: bool
) -> "PreTrainedModel":
r"""Load peft model with unsloth. Used in both training and inference."""
from unsloth import FastLanguageModel # type: ignore
unsloth_kwargs = _get_unsloth_kwargs(config, model_args.adapter_name_or_path[0], model_args)
unsloth_kwargs = _get_unsloth_kwargs(config, model_args.adapter_name_or_path[0], model_args, finetuning_args)
try:
if not is_trainable:
unsloth_kwargs["use_gradient_checkpointing"] = False