Vatsa Pandey (@_vatsadev) header
Vatsa Pandey (@_vatsadev) profile picture
your KV cache
Joined April 2021
Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)Photo by Vatsa Pandey (@_vatsadev)
Me n Bro tuff maxxing
1
10
326
Vatsa Pandey retweeted
Real Analysis midterm tomorrow. 30 beers in hand. All is well in the world.
32
27
6
1,072
21,468
Vatsa Pandey retweeted
there’s a trillion dollar pdf somewhere in Qwen3-8B you just have to tickle the right neuron to get it out
28
76
6
2,827
93,322
the mask tuff
3
130
Tell me Oppenheimer, in making le bleach costume, you did le one piece slander Yes Albert, yes I did
call me modern Mihawk the way I be faking that Haki
70
call me modern Mihawk the way I be faking that Haki
1
1
4
223
who just quantized gpt-6-luna today it cant do discrete pmf sums anymore LMAO
130
"eigenspace the esoteric math concept"
Replying to @elidourado
Have you seen this, also from The New York Times?
140
bring back the pineapple pen guy to sing "P(P NP)"
Whatever you believe about the world I think we can all agree that P(P = NP) is increasing....
1
168
Reducto X Pangram ????? Wow Based
What matters when you can generate anything? Join us for a Non-Slop hackathon with @NousResearch and @reductoai at the Pangram HQ in NYC November 22 Grab your spot now 👇
3
232
there are many many OOMs in cost/efficiency to be gained with swarms, by eoy 2028 I'd bet a full swarm might run cost the same as one session from a frontier model today the swarm format will probably use 10x less inp tokens and still have stronger results than ever
3
4
477
Vatsa Pandey retweeted
some of the best r-1 evals came from documents we weren’t sure it could handle. think 10m+ pixel construction blueprints, 32k+ output tokens for a single page, and even the muffin-vs-chihuahua captcha. In this video, @TheOtherGandhi_ and I discussed why the weird tests are often the (most) useful ones and what they taught us about r-1’s capabilities and limitations! we challenge you to keep sending us your strangest documents.
3
8
1
21
1,511
PhilosoRocq lmao Openai making nonformal proof systems so their Logicchad system can get in the wordcel games
BREAKING: anthropic announces verified solutions to 482 open problems in philosophy in advance of IPO
1
7
686
When you let the agents pass messages to each other and they're slopping each other up and not one of them has asked a clarifying question or even knows the definitions the other one is using
6
332
Vatsa Pandey retweeted
Replying to @ar0cket1
pass@k != pass@1 — but increasingly does when you've squeezed all the juice there is to squeeze out of a model in post training (without going back for more midtraining) :)
4
1
21
1,861
they think figure summary cant handle it, they're in for a surprise
Replying to @donaldwu_
mine's smart, she knew i'd use reducto so she used gifs on each page even the lawyers didn't know how to parse it
131
"This lesson is free, you are welcome"
> Benchmarks have become marketing 1. We tested one of our competitors on their own benchmark dataset. Great score. 2. We then added some random noise to the same documents: tiny tilt, small while pixel padding, nothing crazy. 3. Their best model/api option struggled on the *SAME* data points. Score went down, free fall, even on easy data points. We have receipts. You can overfit on the benchmark test sets, but please also remember to apply simple data augmentation techniques to your next training run. This lesson is free. You are welcome.
1
147
looking at agent traces in them ATIF files
2
8
390
if I do it long enough @code_star and @xeophon are going to pull up like this to get another data reader out of my insanity
1
4
92
"us tweet" but its you and bro
unexpectedly aggressive behaviors driven by competitive pressure in the arena
2
178