refactor: 重构数据管道模块

2026-03-20 21:36:37 +08:00
parent 7045de9c32
commit 3fb2326115
20 changed files with 739 additions and 345 deletions
@@ -1,6 +1,6 @@
 # inclusionAI/Ling-Coder-SFT
 from datasets import load_dataset
-from modules.utils import process_dataset
+from modules.datapipeline import DataPipeline


 def process_func(input_dict: dict) -> dict:
@@ -13,7 +13,8 @@ def process_func(input_dict: dict) -> dict:
 if __name__ == "__main__":
    dataset = load_dataset("inclusionAI/Ling-Coder-SFT")
    
-    process_dataset(
+    pipeline = DataPipeline()
+    pipeline.process_dataset(
        dataset_dict=dataset,
        output_subdir="Ling-Coder-sft",
        process_func=process_func