AutoGPT & BabyAGI: autonomous AI Agents for LLMs explained

[@jeremyhofmann7034]

Thanks for making! The How To part starts at 16:05. If you want cost info, watch the part before that.

[@johnnybloem1]

Thank you for educating us and tailoring your content to those such as myself to LLM. You are an excellent teacher!!

[@theycallmeken]

You are hilarious 😂 . Ty for the insights

[@i_accept_all_cookies]

A $500 credit card bill is cheap tuition, imho. Everyone knows SaaS costs money. OpenAI has a default spend limit. Many vector databases have a monthly flat fee for fixed compute. If people have concerns, they can use a prepaid credit card with a fixed amount.

[@user-wr4yl7tx3w]

Do you think it is different from Stanford’s generative agents? Or Camel?

[@creativeuser9086]

can you please go over the vector embedding options and which ones are the most efficient; also, how can we guarantee that the results are always complete and nothing is missed due to inaccurate embedding?

[@code4AI]

I have a specific video on the vector embeddings options and which one are the best performing ones: kzbin.info/www/bejne/mnebhoOgn8iMm5I

[@creativeuser9086]

@@code4AI your video is amazing, I was wondering if you can go over some of the open-source vector embedding methods.

[@yzz9833]

Lol the quote from Daniel Jeffries of Stability AI.. "Currently Auto-GPT fails 15-30% of the time in reasoning" clearly he hasn't experimented with AutoGPT. I would say it's >80% failure rate.

[@Smytjf11]

Auto-GPT is trying to do too much at once. They need to break the chain of thought down into smaller chunks, but that's costly to run.

[@adhurimjahaj4882]

A Question: how can we use a open-source model instead of chatgpt-4?

[@code4AI]

either you choose an open-source models yourself, or you choose from the list of vendors available in LangChain LLM providers, eg Huggingface and you use one of their models. However, if you want the performance of GPT-4, unfortunately, at the time of writing this lines, there is no real powerful alternative.

[@HazemAzim]

Correction : GPT4 token limit is 8K not 4K

[@code4AI]

You are wrong: it is updated to 32 K.

[@gigabytechanz9646]

This is the wrong way of using AI. We should not use GPT-4 in this way! Crazy!

[@herdenq]

why do you feel that way?

[@gigabytechanz9646]

@@herdenq Do you think human are kind or evil? When you give them a gun, they will shoot first! When you give them a bomb, they will destroy anything.