Merge pull request #69 from langchain-ai/mattf/update-ranking-support

update ranking nim support for nims w/ /v1/models and multiple names

libs/ai-endpoints/langchain_nvidia_ai_endpoints/reranking.py

@@ -27,7 +27,6 @@ class Config:
     _client: _NVIDIAClient = PrivateAttr(_NVIDIAClient)
 
     _default_batch_size: int = 32
-    _deprecated_model: str = "ai-rerank-qa-mistral-4b"
     _default_model_name: str = "nv-rerank-qa-mistral-4b:1"
 
     base_url: str = Field(
@@ -92,7 +91,7 @@ def get_available_models(
     def _rank(self, documents: List[str], query: str) -> List[Ranking]:
         response = self._client.client.get_req(
             payload={
-                "model": "nv-rerank-qa-mistral-4b:1",
+                "model": self.model,
                 "query": {"text": query},
                 "passages": [{"text": passage} for passage in documents],
             },

libs/ai-endpoints/tests/integration_tests/conftest.py

@@ -73,19 +73,11 @@ def get_all_known_models() -> List[Model]:
         metafunc.parametrize("chat_model", models, ids=models)
 
     if "rerank_model" in metafunc.fixturenames:
-        models = ["nv-rerank-qa-mistral-4b:1"]
+        models = [NVIDIARerank._default_model_name]
         if model := metafunc.config.getoption("rerank_model_id"):
             models = [model]
-        # nim-mode reranking does not support model listing via /v1/models endpoint
         if metafunc.config.getoption("all_models"):
-            if mode.get("mode", None) == "nim":
-                models = [model.id for model in NVIDIARerank(**mode).available_models]
-            else:
-                models = [
-                    model.id
-                    for model in get_all_known_models()
-                    if model.model_type == "ranking"
-                ]
+            models = [model.id for model in NVIDIARerank(**mode).available_models]
         metafunc.parametrize("rerank_model", models, ids=models)
 
     if "vlm_model" in metafunc.fixturenames: