Skip to content

Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.

fix(dataset): use supported default QA model (#3965)

xxxxx0ec945fe5df0
diff --git a/dataset/README b/dataset/README
index 6c78bcaa3..8ea8bc559 100644
--- a/dataset/README
+++ b/dataset/README
@@ -62,6 +62,8 @@ python dataset/qa_dataset.py generate \
--output_dir ./dataset/samples
```
+The script uses `gpt-5.6-sol` by default.
+
### 3. View Results
Display the generated QA pairs with their context:
@@ -270,4 +272,4 @@ Feel free to submit issues and enhancement requests!
## License
-MIT License - feel free to use this tool for your research and projects.
\ No newline at end of file
+MIT License - feel free to use this tool for your research and projects.
diff --git a/dataset/README_zh.md b/dataset/README_zh.md
index 2c4e5b0dc..ad1efceba 100644
--- a/dataset/README_zh.md
+++ b/dataset/README_zh.md
@@ -73,6 +73,8 @@ python dataset/qa_dataset.py generate \
--output_dir ./dataset/samples
```
+脚本默认使用 `gpt-5.6-sol`。
+
### 3. 查看结果
展示生成的问答对及其上下文:
@@ -281,4 +283,4 @@ python dataset/qa_dataset.py show \
## 许可证
-MIT许可证 - 可自由用于研究和项目。
\ No newline at end of file
+MIT许可证 - 可自由用于研究和项目。
diff --git a/dataset/qa_dataset.py b/dataset/qa_dataset.py
index df4ee5229..393d848bf 100644
--- a/dataset/qa_dataset.py
+++ b/dataset/qa_dataset.py
@@ -33,6 +33,9 @@ import pandas as pd
import openai
+DEFAULT_QA_MODEL = "gpt-5.6-sol"
+
+
def read_parquet(path):
return pd.read_parquet(path)
@@ -145,7 +148,7 @@ class QAAnsweringSystem:
return "\n\n".join(context_parts)
- def answer_question(self, qid: str, model: str = "gpt-4o-2024-05-13") -> str:
+ def answer_question(self, qid: str, model: str = DEFAULT_QA_MODEL) -> str:
"""
Use OpenAI API to answer question based on qid context
diff --git a/dataset/tests/test_qa_dataset_default_model.py b/dataset/tests/test_qa_dataset_default_model.py
new file mode 100644
index 000000000..d8c3e81a4
--- /dev/null
+++ b/dataset/tests/test_qa_dataset_default_model.py
@@ -0,0 +1,38 @@
+"""Regression test for the QA dataset model default."""
+
+import importlib.util
+from pathlib import Path
+from types import SimpleNamespace
+import unittest
+from unittest.mock import Mock, patch
+
+import pandas as pd
+
+
+SOURCE = Path(__file__).resolve().parents[1] / "qa_dataset.py"
+
+
+class QADatasetDefaultModelTests(unittest.TestCase):
+ def test_default_model_reaches_openai_client(self):
+ spec = importlib.util.spec_from_file_location("qa_dataset_under_test", SOURCE)
+ module = importlib.util.module_from_spec(spec)
+ spec.loader.exec_module(module)
+
+ client = Mock()
+ client.chat.completions.create.return_value = SimpleNamespace(
+ choices=[SimpleNamespace(message=SimpleNamespace(content="answer"))]
+ )
+ queries = pd.DataFrame([{"id": "q1", "text": "question"}])
+ corpus = pd.DataFrame([{"id": "p1", "text": "context"}])
+ qrels = pd.DataFrame([{"qid": "q1", "pid": "p1"}])
+
+ with patch.object(module.openai, "Client", return_value=client):
+ system = module.QAAnsweringSystem(queries, corpus, qrels)
+ self.assertEqual(system.answer_question("q1"), "answer")
+
+ request = client.chat.completions.create.call_args.kwargs
+ self.assertEqual(request["model"], "gpt-5.6-sol")
+
+
+if __name__ == "__main__":
+ unittest.main()