I'm working on a TL;DR bot for Lemmy, powered by GPT-3.5
cross-posted from: https://programming.dev/post/177822
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.
Sounds cool. Looking forward when there will be enough content here to use TL;Dr bot π
It can also summarize links, so itβs already useful even if there are few people posting walls of text
Cool. Who will pay for LLM usage (openai I assume)?
It will be me π
I limited it to 100 summaries per day, so it wonβt cost more than about $20/month in the worst case.
π nice.
i know you've only got 12000 chr but i would love to see this able to read the comments in a thread and categorise them something like '60% of comments are about the erroneous grammar in line 12, 35% are personal anecdotes about the subject, in the remaining 5% the following salient points are raised...'
i'm sure it wouldn't be very accurate a first but everything is a journey and it would be a lot of fun
Yes, there are so many possibilities for GPT-powered bots. I want to implement a couple more.
Are you using OpenAI for this, and if so, is this out of pocket? Secondly, if not, got a particular model to recommend? I wanted to do the same, but can't be bothered to pay for it myself.
Yes, OpenAI and out of pocket. Unfortunately cheaper or self-hosted models are much worse.
Yeah, the few local models I've tried weren't exactly awesome. Glad to know it's not just me then :)
This looks really cool but I can't imagine the cost if it becomes popular
Thatβs why thereβs a daily limit, hopefully Iβll be able to increase it as the API gets cheaper.