| Name | Type | Description |
|---|---|---|
model | Optional[str] | Default: NoneThe model to use for chat. |
nvidia_api_key | Optional[str] | Default: NoneThe API key to use for connecting to the hosted NIM. |
api_key | Optional[str] | Default: None |
base_url | Optional[str] | Default: None |
temperature | Optional[float] | Default: None |
max_completion_tokens | Optional[int] | Default: None |
top_p | Optional[float] | Default: None |
seed | Optional[int] | Default: None |
stop | Optional[Union[str, List[str]]] | Default: None |
default_headers | Optional[Dict[str, str]] | Default: None |
**kwargs | Any | Default: {} |
NVIDIA chat model.
Example:
from langchain_nvidia_ai_endpoints import ChatNVIDIA
model = ChatNVIDIA(model="meta/llama2-70b")
response = model.invoke("Hello")Alternative to nvidia_api_key.
The base URL of the NIM to connect to.
Format for base URL is http://host:port
Sampling temperature in [0, 2].
Maximum number of tokens to generate.
Top-p for distribution sampling in [0, 1].
A seed for deterministic results.
A string or list of strings specifying stop sequences.
Default headers merged into all requests.
Additional parameters passed to the underlying client.