Nuno Campos

icon company @langchain-ai icon location San Francisco icon location Helping people build stuff.

Results 249 comments of Nuno Campos

In a LangGraph app employing Azure OpenAI LLM within multiple nodes, despite setting streaming=False to some of them, the app still streams all LLMs' tokens.

hi, not sure what the question is here, if you call your graph with `astream_events` we try to stream output from all things called inside it. were you expecting to...

Feat/Add @workflow.register_node decorator

I'm not opposed to this as long as we can do it inside the existing add_node method

add easy way to run all examples

You mean run them one at a time, or all of them a bit like a test?

add easy way to run all examples

yea then they should be tests instead, inside the langchain folder. I had thought the examples were to serve the same purpose as the python notebooks

Move code into langchain/src

I tried that, but it didn't fix it

Move code into langchain/src

Done in #124

Faster concurrent embeddings for OpenAI

@vladgolubev Thanks for this. I think part of the reason this wasn't done before is due to openai rate limits. Did you try passing a higher value to the existing...

Faster concurrent embeddings for OpenAI

@vladgolubev we've increased the default value of batchSize, do you want to give it a go on the latest version? OpenAI recommend that batches are sent in the same API...

openai.Completion.acreate(stream=True) raises an Exception

Obviously this is using the newly released version 0.26.0

openai.Completion.acreate(stream=True) raises an Exception

The final issue is in this line https://github.com/openai/openai-python/blob/4fee0da142f58307edb30111ed484d2d41e4811d/openai/api_requestor.py#L297 We exit the `with` statement before the response stream is consumed by the caller, therefore, unless we're using a global ClientSession, the...