delete the default model value from langchain and discard the need fo… (

#24915) - description: I remove the limitation of mandatory existence of `QIANFAN_AK` and default model name which langchain uses cause there is already a default model nama underlying `qianfan` SDK powering langchain component. --------- Co-authored-by: Chester Curme <[email protected]>
langchain-ai · Aug 6, 2024 · 061ed25 · 061ed25
1 parent 293a4a7
commit 061ed25
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 12 deletions.
diff --git a/libs/community/langchain_community/chat_models/baidu_qianfan_endpoint.py b/libs/community/langchain_community/chat_models/baidu_qianfan_endpoint.py
@@ -346,7 +346,9 @@ class Joke(BaseModel):
 
     client: Any  #: :meta private:
 
-    qianfan_ak: SecretStr = Field(alias="api_key")
+    # It could be empty due to the use of Console API
+    # And they're not list here
+    qianfan_ak: Optional[SecretStr] = Field(default=None, alias="api_key")
     """Qianfan API KEY"""
     qianfan_sk: Optional[SecretStr] = Field(default=None, alias="secret_key")
     """Qianfan SECRET KEY"""
@@ -365,13 +367,13 @@ class Joke(BaseModel):
     In the case of other model, passing these params will not affect the result.
     """
 
-    model: str = "ERNIE-Lite-8K"
+    model: Optional[str] = Field(default=None)
     """Model name.
     you could get from https://cloud.baidu.com/doc/WENXINWORKSHOP/s/Nlks5zkzu
     
     preset models are mapping to an endpoint.
     `model` will be ignored if `endpoint` is set.
-    Default is ERNIE-Lite-8K.
+    Default is set by `qianfan` SDK, not here
     """
 
     endpoint: Optional[str] = None
@@ -386,16 +388,12 @@ class Config:
     def validate_environment(cls, values: Dict) -> Dict:
         values["qianfan_ak"] = convert_to_secret_str(
             get_from_dict_or_env(
-                values,
-                ["qianfan_ak", "api_key"],
-                "QIANFAN_AK",
+                values, ["qianfan_ak", "api_key"], "QIANFAN_AK", default=""
             )
         )
         values["qianfan_sk"] = convert_to_secret_str(
             get_from_dict_or_env(
-                values,
-                ["qianfan_sk", "secret_key"],
-                "QIANFAN_SK",
+                values, ["qianfan_sk", "secret_key"], "QIANFAN_SK", default=""
             )
         )
 

diff --git a/libs/community/langchain_community/embeddings/baidu_qianfan_endpoint.py b/libs/community/langchain_community/embeddings/baidu_qianfan_endpoint.py
@@ -55,7 +55,7 @@ class QianfanEmbeddingsEndpoint(BaseModel, Embeddings):
     chunk_size: int = 16
     """Chunk size when multiple texts are input"""
 
-    model: str = "Embedding-V1"
+    model: Optional[str] = Field(default=None)
     """Model name
     you could get from https://cloud.baidu.com/doc/WENXINWORKSHOP/s/Nlks5zkzu
     

diff --git a/libs/community/langchain_community/llms/baidu_qianfan_endpoint.py b/libs/community/langchain_community/llms/baidu_qianfan_endpoint.py
@@ -55,12 +55,14 @@ class QianfanLLMEndpoint(LLM):
     streaming: Optional[bool] = False
     """Whether to stream the results or not."""
 
-    model: str = "ERNIE-Bot-turbo"
+    model: Optional[str] = Field(default=None)
     """Model name. 
     you could get from https://cloud.baidu.com/doc/WENXINWORKSHOP/s/Nlks5zkzu
     
     preset models are mapping to an endpoint.
     `model` will be ignored if `endpoint` is set
+    
+    Default is set by `qianfan` SDK, not here
     """
 
     endpoint: Optional[str] = None

diff --git a/libs/community/tests/integration_tests/chat_models/test_qianfan_endpoint.py b/libs/community/tests/integration_tests/chat_models/test_qianfan_endpoint.py
@@ -306,7 +306,10 @@ def test_functions_call() -> None:
 
 def test_rate_limit() -> None:
     chat = QianfanChatEndpoint(model="ERNIE-Bot", init_kwargs={"query_per_second": 2})  # type: ignore[call-arg]
-    assert chat.client._client._rate_limiter._sync_limiter._query_per_second == 2
+    assert (
+        chat.client._client._rate_limiter._internal_qps_rate_limiter._sync_limiter._query_per_second
+        == 1.8
+    )
     responses = chat.batch(
         [
             [HumanMessage(content="Hello")],