feat：embeding改成支持电信模型

.env

@@ -3,8 +3,8 @@ IS_FIRST_LOAD=True
 CHAT_MODEL_BASE_URL=https://api.siliconflow.cn
 CHAT_MODEL_API_KEY=sk-iyhiltycmrfnhrnbljsgqjrinhbztwdplyvuhfihcdlepole
 CHAT_MODEL_NAME=zai-org/GLM-4.5
-
-EMBEDDING_MODEL_BASE_URL=https://api.siliconflow.cn
+#使用ai中台的模型
+EMBEDDING_MODEL_BASE_URL=http://10.225.128.2:13206/member1/small-model/bge/encode
 EMBEDDING_MODEL_API_KEY=sk-iyhiltycmrfnhrnbljsgqjrinhbztwdplyvuhfihcdlepole
 EMBEDDING_MODEL_NAME=BAAI/bge-m3
 

service/cus_vanna_srevice.py

@@ -270,21 +270,21 @@ class CustomQdrant_VectorStore(Qdrant_VectorStore):
 
         request_body = {
             "model": self.embedding_model_name,
-            "input": data,
+            "sentences": [data],
         }
         request_body.update(kwargs)
 
         response = requests.post(
-            url=f"{self.embedding_api_base}/v1/embeddings",
+            url=f"{self.embedding_api_base}",
             json=request_body,
-            headers={"Authorization": f"Bearer {self.embedding_api_key}"},
+            headers={"Authorization": f"Bearer {self.embedding_api_key}", 'Content-Type': 'application/json'},
         )
         if response.status_code != 200:
             raise RuntimeError(
                 f"Failed to create the embeddings, detail: {_get_error_string(response)}"
             )
         result = response.json()
-        embeddings = [d["embedding"] for d in result["data"]]
+        embeddings = result['embeddings']
         return embeddings[0]
 
 class CustomVanna(CustomQdrant_VectorStore, OpenAICompatibleLLM):