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

refactor adapter hparam

Browse Source 
Former-commit-id: f82aece9ebd6df83a7a005cc7cbbcec07fa6e14d
This commit is contained in:
hiyouga
2023-12-15 20:53:11 +08:00
parent 27ef5b1aa7
commit f902b0d420
21 changed files with 302 additions and 311 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/llmtuner/train/utils.py
View File

@@ -46,7 +46,7 @@ def create_ref_model(
ref_model_args_dict = model_args.to_dict()
ref_model_args_dict.update(dict(
model_name_or_path=finetuning_args.ref_model,
checkpoint_dir=finetuning_args.ref_model_checkpoint,
adapter_name_or_path=finetuning_args.ref_model_adapters,
quantization_bit=finetuning_args.ref_model_quantization_bit
))
ref_model_args = ModelArguments(**ref_model_args_dict)
@@ -96,7 +96,7 @@ def create_reward_model(
reward_model_args_dict = model_args.to_dict()
reward_model_args_dict.update(dict(
model_name_or_path=finetuning_args.reward_model,
checkpoint_dir=finetuning_args.reward_model_checkpoint,
adapter_name_or_path=finetuning_args.reward_model_adapters,
quantization_bit=finetuning_args.reward_model_quantization_bit
))
reward_model_args = ModelArguments(**reward_model_args_dict)