rohan anil

@_arohan_

co-founder, board @coreautoai - and continuing to aspire to understand deep learning.

Joined December 2017

Pinned
rohan anil
@_arohan_
Apr 19
It turns out multi step backpropaganda is better. paper has a beautiful way of improving backpropagation. One iteration cleanly gets us backprop, multiple iterations get us a preconditioned update.
rohan anil
@_arohan_
Apr 19
Replying to @LinYorker @ryu0000000001 and @weijie444
arxiv.org/abs/2106.06199 Same update here
rohan anil
@_arohan_
1h
some days loss curves tell you something today is such a day.
rohan anil
@_arohan_
Aug 17
Model trained on data a company that seized to exist. Would this make the model that trains on it have existential dread?
Abhijay Rana
@abhijaymrana
Aug 17
Wow, seems like Google is buying Spirit Airlines' enterprise data for $10m (outbidding Mercor at $7.5m). Basically includes every internal document, email, workflow, and codebase for a once $6B company. Honestly, $10m for 34 years of operational data really seems like a steal.
rohan anil
@_arohan_
Aug 17
Hmm, It feels like Claude is quantized to fp64 today.
rohan anil
@_arohan_
Aug 17
I feel like right now is one of the best times ever to be a machine learning researcher. The pace of progress and modern tools and compute available to us is unprecedented. With the field moving this quickly, it almost feels irresponsible to not take part or do something else.