
PhD student @berkeley_ai, prev Math/CS undergrad @PKU1898.
Joined April 2025
- Tweets5
- Following53
- Followers42
- Likes6
Great analysis on why looping alone doesn’t solve implicit reasoning
Frontier labs are rumored to be adopting looped Transformers for their potential to unlock implicit reasoning. But can looped Transformers really master it?
We find a bottleneck that keeps looped Transformers from perfect implicit reasoning. Using mechanistic interpretability, we show how looping helps implicit reasoning, why it still struggles to generalize, and how to fix it.
We introduce DiscoLoop: a residual design for looped Transformers that improves implicit reasoning.
🧵 Long thread ahead 👇
1
1
204
Yihan Geng retweeted
The technical report for Goedel-Prover-V2 is out!
📌 SOTA among all open-source theorem provers
⚡ Among the best overall—including closed-source—under small test-time compute
Read it here: arxiv.org/abs/2508.03613
Yihan Geng retweeted
(1/4)🚨 Introducing Goedel-Prover V2 🚨
🔥🔥🔥 The strongest open-source theorem prover to date.
🥇 #1 on PutnamBench: Solves 64 problems—with far less compute.
🧠 New SOTA on MiniF2F:
* 32B model hits 90.4% at Pass@32, beating DeepSeek-Prover-V2-671B’s 82.4%.
* 8B > 671B: Our 8B model matches DeepSeek-671B on MiniF2F.
📚 Leading on MathOlympiadBench (IMO-level problems)
* Solves 73 vs 50 over 671B DeepSeek Prover
🔓 Website: blog.goedel-prover.com/
🔓 Model 32B: huggingface.co/Goedel-LM/Goe…
🔓 Model 8B huggingface.co/Goedel-LM/Goe…
🔓Data and training pipeline will be released soon.
Amazing Collaborators: @sangertang1999 @Lyubh22 @__zrrr__ @juihuichung @thomaszhao1998 @pero733858111 @thiiis_user @EmilyJge @JingruoS5931 @wujiayun12 @GesiJiri68334 @davidjesusacu @KaiyuYang4 @hongzhou__lin @YejinChoinka @danqi_chen @prfsanjeevarora @chijinML
Yihan Geng retweeted
🚨 Easy math, epic fail! 🚨
Our new benchmark, Ineq-Comp, gives formal theorem provers Lean inequalities... then makes tiny tweaks (duplicating variables, squaring terms) that humans handle easily.
Most provers collapse. Simple composition is still surprisingly hard!








