prev: Machine Learning @Uber, @AWS building stuff 🤖 🚀 https://teapawt.net/t.co/URF0lvQQwT (acq) now @ClipMoveAI
Read ➡
Joined January 2017
- Tweets2.2K
- Following210
- Followers3.6K
- Likes1.9K
James Pog retweeted
fk it we r LIVE
launching early access on product hunt RAW
no bots, no weird click farms, just honest work
GO SUPPORT US PLS
producthunt.com/posts/clipmo…
James Pog retweeted
IMO this is the future for a lot of GenAI apps
Automatically achieve better & more personalized LLM apps by:
- Log all LLM outputs
- Log feedback for those outputs
- Use feedback to build datasets
- Use datasets as few shot examples
Want to do this for your app? Shoot me a DM
⚙️How Dosu used LangSmith to achieve a 30% accuracy improvement with no prompt engineering
One of the goals of LangSmith is to help teams set up a data flywheel. By capturing LLM outputs alongside user feedback, we can help developers automatically use that feedback to improve their application
That's exactly what @devstein64 and the @dosu_ai team did. They have GitHub bot that auto-labels your issues and PRs (among other features).
We sat down with them to understand how they used logging, datasets, and our new automations feature to automatically improve their application by 30% purely by capturing user feedback. We then set up a cookbook so ANY LangSmith user can replicate what they did.
Blog: blog.langchain.dev/dosu-lang…
YouTube: youtu.be/tHZtq_pJSGo
Cookbook: docs.smith.langchain.com/mon…
James Pog retweeted
i wish I realized this sooner but if you are scared to build something bc of big competitors, don't be.
After working in FAANG you realize just how corrupt big corps are
ex: can't do anything without 20 person approval
the solo autist in his moms house can change the world.
James Pog retweeted
Sneak peek of the future 👀
(in excel lol)
AI agents now in google sheets 🤖🧙♂️
Built this for myself since I wanted to scrape web to generate personalized emails in bulk from many data points
If you're in google sheets a lot for work, early access below ⬇️
James Pog retweeted
DONE 🚀
I finally got web agents working with a model besides gpt4
GPT-4 is king but too expensive
here's some ways I've learned to make claude2 work at GPT-5 level:
- use the AWS Bedrock version it feels way better compared to anthropic version, IDKY? pointing to diff model? it feels less guardrailsy. @langchain makes it super easy to switch providers and try them out.
- Claude was trained on XML so your prompt should be full of XML ex: <task>do this and answer like <answer></answer></task> don't try to prompt it like its gpt4
- Break instructions into chains/chunks.
Ex of BAD prompt: I want you to write an essay about the statement {{STATEMENT}}, with three topic sentences arguing for the statement, three topic sentences arguing against the statement, and a conclusion at the end.
Ex of GOOD prompt: Please follow these steps:
1. Write three topic sentences arguing for {{STATEMENT}}.
2. Write three topic sentences arguing against {{STATEMENT}}.
3. Write an essay by expanding each topic sentence from Steps 1 and 2, and adding a conclusion to synthesize the arguments. Please enclose the essay in <essay></essay> tags.
Claude2 is def the most underrated model I've tried so far. A tier. LLM Agent Tier list coming soon?
James Pog retweeted
guys woah 👀 LangGraph is fantastic..
best multi-agent tool I've tried so far!
I made 2 agents help me research food for a trip to Seattle using @ExaAILabs and @langchain 🤖
(code + results below ⬇️ )
James Pog retweeted
lol what 🤯
services will charge you big $ just to generate blog posts
Here's how to do it with @langchain for free:
1) Scrape YouTube videos in your niche
2) Download comments + title
3) Prompt @ollama model to generate posts
4) run daily
5) profit!
In <50 lines of code..
