With all these specialized LLMs it’s going to be critical to have a way to connect them together to take advantage of their unique and specialized capabilities! Exciting release from the Gorilla team!

Agree with this…..”The best models will be used the most. Evals are flawed. Power users will figure out the best models for each use case.”

The ability to run open source language models and providing us a reason we NEED to upgrade our machines. ollama is on 🔥

The future of language models but also the future of software. We will have software that knows what language model to use based on the use case. Orchestrating and coordinating the models will allow us to balance run costs and performance for our unique problem domains.

LLM is a dream machine. Hallucinations are a problem the assistant using the LLM needs to handle, the LLM should keep dreaming.

Interested in incorporating video into your LLM app but don’t want to deal with all the complexity? This looks promising 👇

Big believer in the ability to plug and play different models based on use case. Use the right model for the right job.

ChatGPT 🤖

This is insane. Can't wait for an API to integrate this with AI apps I'm developing. The ability to string these models together will enable a new class of application that will produce mind blowing output (as if this on its own isn't mind blowing enough!)

What are the most successful approaches for creating an AI chatbot to represent a person? Assume the individual has a lot of academic writing that I’d like to embed as knowledge and from that knowledge also be able to represent beliefs, sentiments, and opinions in relevant

Speed is everything youtu.be/y9O0XBALCiI?si…

Anthropic is killing it lately. Love these updates.

Exploring different LLMs ability to enrich location records (~9m). As part of the process exploring local LLM enrichment through llama 3.1 on my Mac. Who has a pointer to the fastest way to run llama 3.1 on Mac (8b or 70b)? Currently using ollama.

I wish more of the top model providers had API’s. There is so much opportunity in stringing together models in workflows. If no API, could we at least make the handoffs between providers a single click?