The evolution last weeks:
models ->
chains ->
tools ->
agents ->
BabyAGI ???

This is the way!

https://python.langchain.com/en/latest/use_cases/agents/baby_agi_with_agent.html

#ml #langchain

I think Visual ChatGPT may slipp under the radar for many people.

The "Promt Manager" in the system may be the first of an evolving industry...not kidding, this may be the "brain-of-AIs"!

The single python file implementation is worth some attention.

https://github.com/microsoft/visual-chatgpt/blob/main/visual_chatgpt.py

The hf space does not work :-(

#ml #chatgpt #huggingface

My initial thought after looking at the #GPT-4 event and browsing through the "technical report" ( technical really? most marketing...).
But with a 32k context window, the matrixes in the attention layers will be huge. What GPU and memory resources are needed...and how much power/carbon footprint?

I think this model is outdated, before it is released.

But i am not sure anyone else is training a model with this amount of data, so i guess we are stuck with it for awhile...

#ml #openai

This was a nice experience, multimodal #transformers from #salesforce BLIP-2

Try yourself at #huggingface
https://huggingface.co/spaces/hysts/BLIP2-with-transformers

See image for a idea what is possible.

I see a great business value both for caption of images and Q&A support.

Put a voice interface in front of this and you have an awesome assistant for people with limited vision :-)

#COSMO is now available on #huggingface
and a chat_with_cosmo.py is available on #github. So i could test this one on my laptop.

COSMO was trained with SODA, a dataset with "social commonsense" and is "only" 3B parameters.

It is not trained to answer questions like #chatgpt , so have right expectations.

But it is really a nice companion...but this conversation got a bit scary 😂

Try yourself:
https://github.com/skywalker023/sodaverse

#ml #ai