Whether to support streaming output？

[zhenxinxu]

No description provided.

[msoedov]

Short answer - No.

from langchain.chat_models import ChatOpenAI
from langchain.schema import (
    HumanMessage,
)


from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
chat = ChatOpenAI(streaming=True, callbacks=[StreamingStdOutCallbackHandler()], temperature=0)
resp = chat([HumanMessage(content="Write me a song about sparkling water.")])

Supporting streaming is possible, but right now, it does not fit the most common 99% of scenarios for this library. Let me know if you believe otherwise