Files
langchain/libs/partners/huggingface/tests/integration_tests/test_chat_models.py

23 lines
751 B
Python

from langchain_core.messages import AIMessageChunk
from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
def test_stream_usage() -> None:
"""Test we are able to configure stream options on models that require it."""
llm = HuggingFaceEndpoint( # type: ignore[call-arg] # (model is inferred in class)
repo_id="google/gemma-3-27b-it",
task="conversational",
provider="nebius",
)
model = ChatHuggingFace(llm=llm, stream_usage=True)
full: AIMessageChunk | None = None
for chunk in model.stream("hello"):
assert isinstance(chunk, AIMessageChunk)
full = chunk if full is None else full + chunk
assert isinstance(full, AIMessageChunk)
assert full.usage_metadata