Boris Dayma 🖍️
I like the new JAX/Flax API for large models.

I'll be playing with it in my CLIP repo and experiment with how fast we can train a small CLIP model.
tweet

Linus (●ᴗ●)
I love these DM’s so productive. Thank you for the discourse, first sending this then blocking me. 👍🏻 https://t.co/5ZQU8OP4Sr
tweet

Daily AI Papers
Parallel Context Windows Improve In-Context Learning of Large Language Models

https://t.co/0rzIMsVIsu

Large Language Models (LLMs) are handicapped by their limited context windows. We present a method that alleviates...

🧵 👇 https://t.co/bx6ASdJWub
tweet

Convergence Boy
When I asked ChatGPT of crazy ideas to 3D-Print for Christmas, it gave me (after several insists of being "crazier") only "musical instruments for animals", "custom clothes for plants" and "jewelry for pets".
It didn't get anywhere near printing a skull of my kid from her CT scan
tweet

Vova “words are a motherfucker” Zakharov
Damn, why does the bloated, verbose HTML looks better than the logical, succinct @pug_js?😭

I’m going to hate myself for this, but I think i’m going back to the roots.

(Also, nevermind the excrements.) https://t.co/s69IdWmSJU
tweet

Ethan Mollick
What happened when Wellesley College switched to pass/fail grading for first semester students?

It caused students future grades to drop by 23% of a standard deviation, apparently due to lower effort and preparation by students in ungraded courses. https://t.co/LirH5pXqOG https://t.co/sTmuX9nvfM
tweet

Bojan Tunguz
There are three types of people:

1) i++
2) i+=1
3) i=i+1
tweet

Sanskar Tiwari
11 more to go! #buildinpublic https://t.co/hmroJdHY6e
tweet

AI Breakfast
2022: 🤖 < 🧠

2023: 🤖 > 🧠
tweet

Bojan Tunguz
What is something that you can say during sex, but also while your ML model is training?
tweet