BTech/CS IIT, PhD/ML CMU. Adjunct Prof @CMU. Building Langroid -- The easiest way to develop LLM Apps: https://xitter.cf/t.co/xSpEFIaVLv see me on bsky
Pittsburgh, PA
Joined May 2007
- Tweets700
- Following477
- Followers696
- Likes1.5K
Pinned Tweet
[1/5] To the growing list of silly-sounding but useful #LLM ideas (take a deep breath, take a step back, shouting, pleading), add another one: 👉Don't ask an LLM to "parrot" out relevant sentences from a document. Instead, NUMBER the sentences and ask which ones are relevant!
prasad chalasani retweeted
Langroid agents now speak UI ✨
Community developer - Muhammad Asad Ishfaq just shipped the Langroid (github.com/langroid/langroid) integration for AG-UI - the open protocol that standardizes how agents connect to user interfaces.
@pchalasani
prasad chalasani retweeted
Our work on Pharmacovigilance using multi-agent LLM powered system. This was joint work with Duke medical and especially proud of the students for pulling this off. Blog post by @pchalasani about the paper.
Paper is going to appear in MLHC 2024.
Paper: arxiv.org/abs/2408.01869
Blog:
langroid.github.io/langroid/…
prasad chalasani retweeted
🚀 Excited to present our work at MLHC 2024 tomorrow in beautiful Toronto 🇨🇦
Join us as we introduce MALADE—the first effective explainable multi-agent LLM system for extracting Adverse Drug Events (ADEs) from FDA drug labels and drug prescription data.
prasad chalasani retweeted
A big milestone for @langroid_ai . We crossed 1k stars.
As you know the project lead is the amazing @pchalasani , and we have some amazing contributors. Several companies have started using Langroid and are very impressed with the elegance and structure of the framework.
We want to keep building the community.
Try Langroid here: github.com/langroid/langroid
Let us keep the community growing.
prasad chalasani retweeted
Langroid: Complex AI Agents For FREE Better Than Langflow & Flowise! Rivet 2.0? - youtu.be/7FajaWwKAsQ
[1/5] The @OpenAI Assistants API is very interesting.
There is a new programming paradigm emerging, where these assistants are primitives, and a key challenge is:
How can you have assistants work together to solve a task?
[4/5]
It builds up to a two-agent information-extraction task, where:
🤖 the Retriever Agent answers questions from a lease doc
🤖 the Extractor Agent is told to use a function-call to extract structured info about a lease it has no access to, so it asks the Retriever Agent
1
1
128
[5/5]
Be sure to check out the Langroid repo:
github.com/langroid/langroid
And don't forget to star it! ⭐️
We just crossed 20,000 lines of code! 🚀💥
2
138
[5/5] On this example, the Langroid numbering approach is:
😎 22x faster and 36% cheaper 💥🚀
than LangChain's "parrot" approach (name not inspired by the logo I swear 😀 )
See Langroid repo (and don't forget to star it!) github.com/langroid/langroid
#LLM #rag #AI #NLP
1
233
New on the Langroid blog: LLMs are fundamentally "completion" models. This post is a refresher on chat-completion and some interesting details on how chat-completion endpoints are implemented in practice. langroid.substack.com/p/llm-…
98
New on the Langroid blog ! Why Local LLMs are exciting, and how to use them with Langroid.
open.substack.com/pub/langro… #LLM #AI
Announcing the Langroid Blog & Newsletter.
Langroid simplifies #LLM app development using multi-#agent programming.
Posts will be about Langroid, specific agents/applications we are building, as well as on general LLM topics.
open.substack.com/pub/langro…
[1/9] Curious how an #LLM can be used in #marketing ?
We just put out an audience #targeting demo using Langroid, our open-source multi-agent framework that simplifies building LLM applications: github.com/langroid/langroid
[8/9]
Langroid makes it intuitive and simple to build an LLM-powered system organized around agents, each responsible for a different task. In less than a day we built a 3-agent system to automate this task. Read more here: langroid.github.io/langroid/…
1
100