From fd99b9d852a19f57d4f8951e6f0a424c0edf9694 Mon Sep 17 00:00:00 2001
From: ViperEkura <3081035982@qq.com>
Date: Thu, 4 Jun 2026 14:27:04 +0800
Subject: [PATCH] feat: add OpenHermes-2.5 SFT export script

---
 .../supervised_finetuning/sft_openhermes.py   | 22 +++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 scripts/supervised_finetuning/sft_openhermes.py

diff --git a/scripts/supervised_finetuning/sft_openhermes.py b/scripts/supervised_finetuning/sft_openhermes.py
new file mode 100644
index 0000000..b84d79c
--- /dev/null
+++ b/scripts/supervised_finetuning/sft_openhermes.py
@@ -0,0 +1,22 @@
+from datasets import load_dataset
+from pipeline import export_dataset
+
+
+def process_func(input_dict: dict):
+    conversations = input_dict["conversations"]
+    examples = []
+    for i in range(0, len(conversations) - 1, 2):
+        user_msg = conversations[i]["value"]
+        assistant_msg = conversations[i + 1]["value"]
+        examples.append({"query": user_msg, "response": assistant_msg})
+    return examples
+
+
+if __name__ == "__main__":
+    dataset = load_dataset("teknium/OpenHermes-2.5")
+    export_dataset(
+        dataset=dataset["train"],
+        output_dir="./dataset",
+        output_prefix="OpenHermes-2.5",
+        process_func=process_func,
+    )