Thank you for this video! For now, I think tools have replaced Function calling. There are 2 tools introduced code interpreter and file retriever. In future, they will introduce more tools i guess. What do you think is the difference between langchain agents and assistant APIs?

@codingcrashcourses8533

5 ай бұрын

Function calling is also just a tool, too. The difference between agents and assistants is the level of autonomy. My guess is that in the future, an Assistant is part of an Agent system :)

Thank you for the content! Let me enjoy the oportunity to ask you about the Warning of the news OpenAI embeddings models, "Using cl100k_base encoding" Do you know how to pass this encoding model as parameter to eliminate this warning? I already tried to pass the parameter 'tiktoken_model_name' with 'cl100k_base', but doesnt worked.

@codingcrashcourses8533

5 ай бұрын

In short - you can not and you probably should not do that. OpenAI maps their models to the encoder. They just did not update the mapping. But it works, the new models use cl100k_base :)

@matheussa8293

5 ай бұрын

@@codingcrashcourses8533 yeah i saw in the openai docs that the embeddings models, both v2 and v3 uses cl100k_base

@matheussa8293

5 ай бұрын

@@codingcrashcourses8533 yeah i saw in the openai docs that the embeddings models, both v2 and v3 uses cl100k_base, and worked fine