refactor(utils): 将 pre_tarin_process 和 sft_process 函数合并为 process_dataset

2025-07-16 12:17:58 +08:00
parent f64a1a15ee
commit faf9f81e7c
6 changed files with 21 additions and 50 deletions
@@ -1,5 +1,5 @@
 from datasets import load_dataset
-from utils import pre_tarin_process
+from utils import process_dataset

 if __name__ == "__main__":
    max_chunk_num = 10
@@ -10,7 +10,7 @@ if __name__ == "__main__":
        data_files={"train": [f"data/0000{i}.parquet" for i in range(5)]}
    )
    
-    pre_tarin_process(
+    process_dataset(
        dataset_dict=dataset,
        output_subdir="chinese-wiki",
        max_chunk_num=max_chunk_num,