From 897dcaf386499e88a2ed5da752c3af9d8182b750 Mon Sep 17 00:00:00 2001
From: 997146918 <997146918@qq.com>
Date: Mon, 11 Aug 2025 14:00:55 +0800
Subject: [PATCH] =?UTF-8?q?=E5=A4=84=E7=90=86=E6=95=B0=E6=8D=AE=E7=A9=BA?=
 =?UTF-8?q?=E7=9A=84=E6=83=85=E5=86=B5?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 AITrain/example.py | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/AITrain/example.py b/AITrain/example.py
index e5ad617..6767aa6 100644
--- a/AITrain/example.py
+++ b/AITrain/example.py
@@ -40,9 +40,15 @@ def main():
             try:
                 response = json.loads(response)
                 for item in response:
-                    with open(f'{file_name}.jsonl', 'a', encoding='utf-8') as f:
-                        json.dump(item, f, ensure_ascii=False)
-                        f.write('\n')
+                    # 数据质量检查：过滤空的instruction或output
+                    if (item.get('instruction', '').strip() and 
+                        item.get('output', '').strip() and
+                        item.get('character', '').strip()):
+                        with open(f'{file_name}.jsonl', 'a', encoding='utf-8') as f:
+                            json.dump(item, f, ensure_ascii=False)
+                            f.write('\n')
+                    else:
+                        print(f"跳过空字段数据: instruction='{item.get('instruction', '')}', output='{item.get('output', '')}', character='{item.get('character', '')}'")
             except Exception as e:
                 print(f"解析错误: {e}")
                 print(f"原始响应: {repr(response[:200])}")  # 打印前200字符用于调试