Check out our new blog post on talking like a graph!

Talk Like a Graph or Let Your Graph Do the Talking 🧠🕸️
New paper alert 🚨: we show how GNNs can effectively encode graph-structured data for LLMs: arxiv.org/abs/2402.05862

𝐋𝐞𝐭 𝐘𝐨𝐮𝐫 𝐆𝐫𝐚𝐩𝐡 𝐃𝐨 𝐭𝐡𝐞 𝐓𝐚𝐥𝐤𝐢𝐧𝐠: 𝐄𝐧𝐜𝐨𝐝𝐢𝐧𝐠 𝐒𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞𝐝 𝐃𝐚𝐭𝐚 𝐟𝐨𝐫 𝐋𝐋𝐌𝐬

Don't know what to do with your graphs in 2024? Shove them to an LLM, of course, and let LLM figure out what to do!
arxiv.org/abs/2402.05862
Short thread (1/5):

🚨Get more out of your #LLM with a simple trick!

Few-shotting an #LLM with manually written or randomly selected demonstrations may not be optimal. Read our latest survey to learn about how demonstration retrieval can give a bump in performance.

Do you know a hospital in Canada that would quickly do a liver transplant in a humanitarian emergency at a reasonable price? If so, please contact Nazanin.

And please retweet.

Geometry indeed seems to be a great direction for testing and enhancing machine reasoning. It requires abstract thinking & arithmetic & logic verification. Such capabilities are obviously important for many many tasks.

A geometric interpretation is worth 1000 proof lines. Continuing my love on using AI for Math, we ask how well does Vision-Lang models do on Geometry which mixes calc with geometric info in figure. We propose GeomVerse: a synthetic Geometry dataset. Led by brilliant Mehran Kazemi

This work on discovering new lower bounds by tabu-search and Alpha-Zero will be presented by Gheorghe Comanici at #NeurIPS2023 MathAI workshop today. Have a look to discuss how we can advance some of this work in maths using AI.

i'm co-hosting a student researcher at Google AI w/ Deepak Ramachandran!

We're looking for folks interested in AI assisted creativity, re-imagining interaction as a co-creation process. Background in HCI or human exps + experience with generative models would be a good fit 🧵👇