@shreyaskapur

PhD student @berkeley_ai. Prev. undergrad @MIT, intern @Waymo @GoogleDeepMind

 Berkeley, CA
shreyaskapur.com
Joined June 2012
  • Tweets53
  • Following202
  • Followers2.5K
9Media
My first PhD paper!🎉We learn *diffusion* models for code generation that learn to directly *edit* syntax trees of programs. The result is a system that can incrementally write code, see the execution output, and debug it. 🧵1/n
111
580
125
5,325
742,831
wow
Sergey Levine
@svlevine22 Apr 2025
π-0.5 is here, and it can generalize to new homes! Some fun experiments with my colleagues at @physical_int, introducing π-0.5 (“pi oh five”). Our new VLA can put dishes in the sink, clean up spills and do all this in homes that it was not trained in🧵👇
3
877
I've been waiting 10 years to make this.
187
489
69
7,703
785,378
Built with Google Gemini Flash 2.0 Image generation :D
9
8
1
293
27,298
Can LLMs do lateral thinking puzzles? I tested a bunch of language models on questions from @lateralcast and the #OnlyConnect gameshow! (1/2) 🧵
1
7
2,951
I wrote up the full results on my blog, shreyaskapur.com/blogs/later… alongside example outputs from models. (2/2)
3
1,481
Shreyas Kapur retweeted
Jiahai Feng@feng_jiahai17 Dec 2024
LMs can generalize to implications of facts they are finetuned on. But what mechanisms enable this, and how are these mechanisms learned in pretraining? We develop conceptual and empirical tools for studying these qns. 🧵
5
21
2
148
24,754
Come check out my tree diffusion poster at the system 2 reasoning at scale workshop at NeurIPS!
Shalev
@Shalev_lif15 Dec 2024
Best poster moment at #NeurIPS2024
1
33
2,819
Shreyas Kapur retweeted
Luke Bailey
@LukeBailey18113 Dec 2024
Can interpretability help defend LLMs? We find we can reshape activations while preserving a model’s behavior. This lets us attack latent-space defenses, from SAEs and probes to Circuit Breakers. We can attack so precisely that we make a harmfulness probe output this QR code. 🧵
11
82
19
371
59,093
I'll be at NeurIPS, let me know if you want to catch up or chat about program synthesis, world models, neurosymbolic, search, probabilistic programming, or mourning the loss of King Da Ka.
1
1
14
1,203
I am currently holding my dad's cryopreserved brain tumor samples in hopes of creating a personalized vaccine for immunotherapy. However, there are some critical and time-sensitive questions in the attached post: xitter.cf/tejasdkulkarni/status/… This is time-sensitive so would appreciate any DMs/RTs.
Tejas Kulkarni
@tejasdkulkarni15 Jun 2024

Creating a personalized vaccine for my dad's high-grade brain cancer.

Article · @tejasdkulkarni
15
57
3
162
43,991
I had a lot of fun working on this. I didn't believe that a chess playing neural net could learn to do look-ahead just in its weights, so I was definitely the non-believer in this project.
Erik Jenner@jenner_erik4 Jun 2024
♟️Do chess-playing neural nets rely purely on simple heuristics? Or do they implement algorithms involving *look-ahead* in a single forward pass? We find clear evidence of 2-turn look-ahead in a chess-playing network, using techniques from mechanistic interpretability! 🧵
2
12
1
205
22,623
These languages are small, and we only show this approach on a fairly narrow inverse-graphics task. In the future, we hope to show that this approach may potentially work more generally with languages with loops and variables. 8/n
1
1
117
15,908
We managed to get part of our project running in the browser, Website🌎: tree-diffusion.github.io Paper📄: arxiv.org/abs/2405.20519 Code🖥️: github.com/revalo/tree-diffu… Thanks for my wonderful collaborator @jenner_erik, and advisor Stuart Russell! n/n 🧵
6
11
241
15,727
Load more