Skip to content

The agent engineering platform.

partners/HuggingFacePipeline[stream]: Change to use `pipeline` instead of `pipeline.model.generate` in stream() (#26531)

Hyejun An6227396e20aa
diff --git a/libs/partners/huggingface/langchain_huggingface/llms/huggingface_pipeline.py b/libs/partners/huggingface/langchain_huggingface/llms/huggingface_pipeline.py
index 9c8d5f7145..b1261bcfc4 100644
--- a/libs/partners/huggingface/langchain_huggingface/llms/huggingface_pipeline.py
+++ b/libs/partners/huggingface/langchain_huggingface/llms/huggingface_pipeline.py
@@ -349,7 +349,6 @@ class HuggingFacePipeline(BaseLLM):
stopping_criteria = StoppingCriteriaList([StopOnTokens()])
- inputs = self.pipeline.tokenizer(prompt, return_tensors="pt")
streamer = TextIteratorStreamer(
self.pipeline.tokenizer,
timeout=60.0,
@@ -357,12 +356,12 @@ class HuggingFacePipeline(BaseLLM):
skip_special_tokens=True,
)
generation_kwargs = dict(
- inputs,
+ text_inputs=prompt,
streamer=streamer,
stopping_criteria=stopping_criteria,
**pipeline_kwargs,
)
- t1 = Thread(target=self.pipeline.model.generate, kwargs=generation_kwargs)
+ t1 = Thread(target=self.pipeline, kwargs=generation_kwargs)
t1.start()
for char in streamer: