diff --git a/belle_sft.py b/sft_belle.py
similarity index 100%
rename from belle_sft.py
rename to sft_belle.py
diff --git a/chinese-instruct.py b/sft_chinese_instruct.py
similarity index 100%
rename from chinese-instruct.py
rename to sft_chinese_instruct.py
diff --git a/sft_coder.py b/sft_coder.py
new file mode 100644
index 0000000..3099f16
--- /dev/null
+++ b/sft_coder.py
@@ -0,0 +1,21 @@
+# inclusionAI/Ling-Coder-SFT
+from datasets import load_dataset
+from utils import process_dataset
+
+
+def process_func(input_dict: dict) -> dict:
+    msg = input_dict["messages"]
+    query = msg[0]["content"]
+    history = msg[1]["content"]
+    return {"query": query, "response": history}
+
+
+if __name__ == "__main__":
+    dataset = load_dataset("inclusionAI/Ling-Coder-SFT")
+    
+    process_dataset(
+        dataset_dict=dataset,
+        output_subdir="Ling-Coder-SFT",
+        process_func=process_func
+    )
+    
\ No newline at end of file
diff --git a/utils.py b/utils.py
index c76247b..86906b8 100644
--- a/utils.py
+++ b/utils.py
@@ -102,7 +102,7 @@ def dump_pkl_files(
             sequence = torch.cat(package[key])
             output_package[key] = sequence
          
-        with open(out_file_path, "w") as f:
+        with open(out_file_path, "wb") as f:
             pkl.dump(output_package, f)