mirror of
https://github.com/hwchase17/langchain.git
synced 2025-08-26 21:11:25 +00:00
**Description:** The test_sparkllm.py can reproduce this issue. https://github.com/langchain-ai/langchain/blob/master/libs/community/tests/integration_tests/chat_models/test_sparkllm.py#L66 ``` Testing started at 18:27 ... Launching pytest with arguments test_sparkllm.py::test_chat_spark_llm --no-header --no-summary -q in /Users/zhanglei/Work/github/langchain/libs/community/tests/integration_tests/chat_models ============================= test session starts ============================== collecting ... collected 1 item test_sparkllm.py::test_chat_spark_llm ============================== 1 failed in 0.45s =============================== FAILED [100%] tests/integration_tests/chat_models/test_sparkllm.py:65 (test_chat_spark_llm) def test_chat_spark_llm() -> None: > chat = ChatSparkLLM( spark_app_id="your spark_app_id", spark_api_key="your spark_api_key", spark_api_secret="your spark_api_secret", ) # type: ignore[call-arg] test_sparkllm.py:67: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ../../../../core/langchain_core/load/serializable.py:111: in __init__ super().__init__(*args, **kwargs) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ cls = <class 'langchain_community.chat_models.sparkllm.ChatSparkLLM'> values = {'spark_api_key': 'your spark_api_key', 'spark_api_secret': 'your spark_api_secret', 'spark_api_url': 'wss://spark-api.xf-yun.com/v3.5/chat', 'spark_app_id': 'your spark_app_id', ...} @model_validator(mode="before") @classmethod def validate_environment(cls, values: Dict) -> Any: values["spark_app_id"] = get_from_dict_or_env( values, ["spark_app_id", "app_id"], "IFLYTEK_SPARK_APP_ID", ) values["spark_api_key"] = get_from_dict_or_env( values, ["spark_api_key", "api_key"], "IFLYTEK_SPARK_API_KEY", ) values["spark_api_secret"] = get_from_dict_or_env( values, ["spark_api_secret", "api_secret"], "IFLYTEK_SPARK_API_SECRET", ) values["spark_api_url"] = get_from_dict_or_env( values, "spark_api_url", "IFLYTEK_SPARK_API_URL", SPARK_API_URL, ) values["spark_llm_domain"] = get_from_dict_or_env( values, "spark_llm_domain", "IFLYTEK_SPARK_LLM_DOMAIN", SPARK_LLM_DOMAIN, ) # put extra params into model_kwargs default_values = { name: field.default for name, field in get_fields(cls).items() if field.default is not None } > values["model_kwargs"]["temperature"] = default_values.get("temperature") E KeyError: 'model_kwargs' ../../../langchain_community/chat_models/sparkllm.py:368: KeyError ``` I found that when upgrading to Pydantic v2, @root_validator was changed to @model_validator. When a class declares multiple @model_validator(model=before), the execution order in V1 and V2 is opposite. This is the reason for ChatSparkLLM's failure. The correct execution order is to execute build_extra first. https://github.com/langchain-ai/langchain/blob/langchain%3D%3D0.2.16/libs/community/langchain_community/chat_models/sparkllm.py#L302 And then execute validate_environment. https://github.com/langchain-ai/langchain/blob/langchain%3D%3D0.2.16/libs/community/langchain_community/chat_models/sparkllm.py#L329 The Pydantic community also discusses it, but there hasn't been a conclusion yet. https://github.com/pydantic/pydantic/discussions/7434 **Issus:** #27416 **Twitter handle:** coolbeevip --------- Co-authored-by: vbarda <vadym@langchain.dev> |
||
---|---|---|
.. | ||
__init__.py | ||
anthropic.py | ||
anyscale.py | ||
azure_openai.py | ||
azureml_endpoint.py | ||
baichuan.py | ||
baidu_qianfan_endpoint.py | ||
bedrock.py | ||
cohere.py | ||
coze.py | ||
dappier.py | ||
databricks.py | ||
deepinfra.py | ||
edenai.py | ||
ernie.py | ||
everlyai.py | ||
fake.py | ||
fireworks.py | ||
friendli.py | ||
gigachat.py | ||
google_palm.py | ||
gpt_router.py | ||
huggingface.py | ||
human.py | ||
hunyuan.py | ||
javelin_ai_gateway.py | ||
jinachat.py | ||
kinetica.py | ||
konko.py | ||
litellm_router.py | ||
litellm.py | ||
llama_edge.py | ||
llamacpp.py | ||
maritalk.py | ||
meta.py | ||
minimax.py | ||
mlflow_ai_gateway.py | ||
mlflow.py | ||
mlx.py | ||
moonshot.py | ||
oci_data_science.py | ||
oci_generative_ai.py | ||
octoai.py | ||
ollama.py | ||
openai.py | ||
pai_eas_endpoint.py | ||
perplexity.py | ||
premai.py | ||
promptlayer_openai.py | ||
sambanova.py | ||
snowflake.py | ||
solar.py | ||
sparkllm.py | ||
symblai_nebula.py | ||
tongyi.py | ||
vertexai.py | ||
volcengine_maas.py | ||
yandex.py | ||
yi.py | ||
yuan2.py | ||
zhipuai.py |