I'm working on a TL;DR bot for Lemmy, powered by GPT-3.5
I'm working on a TL;DR bot for Lemmy, powered by GPT-3.5
It's coming along nicely, I hope I'll be able to release it in the next few days.
Screenshot:
How It Works:
I am a bot that generates summaries of Lemmy comments and posts.
- Just mention me in a comment or post, and I will generate a summary for you.
- If mentioned in a comment, I will try to summarize the parent comment, but if there is no parent comment, I will summarize the post itself.
- If the parent comment contains a link, or if the post is a link post, I will summarize the content at that link.
- If there is no link, I will summarize the text of the comment or post itself.
Extra Info in Comments:
Prompt Injection:
Of course it's really easy (but mostly harmless) to break it using prompt injection:
It will only be available in communities that explicitly allow it. I hope it will be useful, I'm generally very satisfied with the quality of the summaries.
You're viewing a single thread.
I'm always curious about using GPT like that. Does it cost money to send requests like this into GPT?
It does unfortunately, see here:
I limited it to 100 summaries / day, which adds up to about $20 (USD) per month if the input is 3000 tokens long and the answer is 1000.
Using it for personal things (I buildt a personal assistant chatbot for myself) is very cheap. But if you use it in anything public, it can get expensive quickly.
Have you considered using a self-hosted instance of GPT4All? It's not as powerful, but for something like summarizing an article it could be plenty - And importantly, much, much cheaper.
How does GPT4ALL work exactly? Surely its no where near the quality of real GPT4.
If i check on https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard The 13B model is below some of the llama models. At this point Falcom 40B is what you want to have the best quality llm running locally
It's just an open source LLM that you can run on your own hardware; I haven't looked into it a ton tbh - but if I saw $20/month for 100 requests/day I'd be immediately looking for a way to run it on my own hardware
Its kinda debatable, while yes I wouldnt wanna pay that either, ive been following the local working llms, Gpt4all stroke me as not bad but not all that special or amazing (compared to 2021 there all magic though) the naming seems a very little bit misleading with gpt4 as the world most advanced known model. All the models on tbe huggingface page i send work can work locally but at best there gpt-3 level competitors.
I havenโt yet looked into it, but the screencast on its website looks really promising! I have a lot on my plate right now so I think Iโll release it first with the GPT-3.5 integration, but Iโll definitely try GPT4All later!