423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-15 11:20:35 +08:00
Code Issues Packages Projects Releases Wiki Activity

implement efficient packing without cross-contamination attention

Browse Source
This commit is contained in:
ancv
2024-06-12 11:56:01 +07:00
parent 972ec9c668
commit b2c367bc61
9 changed files with 287 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llamafactory/train/kto/workflow.py
View File

@@ -24,7 +24,7 @@ def run_kto(
):
tokenizer_module = load_tokenizer(model_args)
tokenizer = tokenizer_module["tokenizer"]
dataset = get_dataset(model_args, data_args, training_args, stage="kto", **tokenizer_module)
dataset = get_dataset(model_args, data_args, training_args, finetuning_args, stage="kto", **tokenizer_module)
model = load_model(tokenizer, model_args, finetuning_args, training_args.do_train)
data_collator = KTODataCollatorWithPadding(