config: set default judge_model=gpt-5, embedding_model=text-embedding-3-small

gpt-5.4/5.5/5.2/5.4-mini/5.4-nano are incompatible with RAGAS 0.4.3
because they require max_completion_tokens instead of max_tokens.
gpt-5 / gpt-4.1 support max_tokens and json_object mode required by RAGAS.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

.env.example

@@ -8,8 +8,10 @@ OPENAI_BASE_URL=http://6.86.80.4:30080/v1
 OPENAI_TIMEOUT_SECONDS=180
 
 # 默认评测模型（可在场景 YAML 或 Web 控制台 LLM 配置中覆盖）
-RAGAS_JUDGE_MODEL=deepseek-v4-flash
-RAGAS_EMBEDDING_MODEL=text-embedding-v3
+# RAGAS_JUDGE_MODEL 需支持 max_tokens + json_object（gpt-5、gpt-4.1、gpt-4o 等）
+# 注意：gpt-5.4/5.5/5.2 系列不支持 max_tokens，与 RAGAS 0.4.3 不兼容
+RAGAS_JUDGE_MODEL=gpt-5
+RAGAS_EMBEDDING_MODEL=text-embedding-3-small
 
 # 评估并发控制（启用 7 个指标时建议 RAGAS_METRIC_TIMEOUT_SECONDS=300）
 BATCH_SIZE=8

rag_eval/settings.py

@@ -21,9 +21,9 @@ class EvaluationSettings(BaseSettings):
 
     openai_api_key: str | None = Field(default=None, alias="OPENAI_API_KEY")
     openai_base_url: str = Field(default="http://6.86.80.4:30080/v1", alias="OPENAI_BASE_URL")
-    ragas_judge_model: str = Field(default="deepseek-v4-flash", alias="RAGAS_JUDGE_MODEL")
+    ragas_judge_model: str = Field(default="gpt-5", alias="RAGAS_JUDGE_MODEL")
     ragas_embedding_model: str = Field(
-        default="text-embedding-v3",
+        default="text-embedding-3-small",
         alias="RAGAS_EMBEDDING_MODEL",
     )
     openai_timeout_seconds: float = Field(default=30.0, alias="OPENAI_TIMEOUT_SECONDS")

webapp/api/score.py

@@ -91,7 +91,7 @@ def score_sample(
     - `semantic_similarity` — 回答与参考答案的语义相似度（需 ground_truth）
 
     **推荐模型配置**：
-    - `judge_model`: `gpt-5.4`
+    - `judge_model`: `gpt-5`
     - `embedding_model`: `text-embedding-3-small`
 
     **鉴权**：若 `.env` 中配置了 `SCORE_API_TOKEN`，需在请求头携带

webapp/models.py

@@ -420,7 +420,7 @@ class ScoreRequest(BaseModel):
                     "context_recall",
                     "context_precision",
                 ],
-                "judge_model": "gpt-5.4",
+                "judge_model": "gpt-5",
                 "embedding_model": "text-embedding-3-small",
             }
         }