Dmitry Rybin
@DmitryRybin1
Followers
1K
Following
14K
Media
34
Statuses
369
PhD at CUHK || ML for Math, Search, Planning || Grand First Prize at IMC || 人工智能+数学
Central & Western District
Joined May 2022
We discovered faster way to compute product of matrix by its transpose!. This has profound implications for data analysis, chip design, wireless communication, and LLM training!. paper: The algorithm is based on the following discovery: we can compute
55
587
4K
I observed this too. For integer initializations and certain sizes Strassen-like algorithms for matmul drastically outperformed baselines. That’s why I always test on random normal data.
0
0
9
Such data should be very useful. A bit confusing to me why Gemini 2.5 is almost human expert level. Perhaps stratification by problem difficulty should reveal more. Tbh never seen a case of a wrong human judging in Russian National Math Olympiads.
Thrilled to share a major step forward for AI for mathematical proof generation! . We are releasing the Open Proof Corpus: the largest ever public collection of human-annotated LLM-generated math proofs, and a large-scale study over this dataset!
1
0
6
RT @yushun_zzz: Would like to share some recent numerical results, which might draw interest from the theory community. We visualized >100….
0
6
0
Very insightful talk by Demis Hassabis at IAS! I didn’t know DeepMind is working on room temperature superconductors!.
0
0
4
Big kudos to Vladimir @shitov_happens for implementing our new X*X^T algorithm in CUDA with no extra memory! . I tried and it is already faster than cublas syrk/gemm in some setups:. 7% faster for 16k x 16k matrices in FP32 on RTX 4090. And comparable in other:. 1-2% slower than.
We discovered faster way to compute product of matrix by its transpose!. This has profound implications for data analysis, chip design, wireless communication, and LLM training!. paper: The algorithm is based on the following discovery: we can compute
3
4
35
Google DeepMind has all the tools, like AlphaProof, to start a new era of scientific LLM benchmarking - test LLMs on open math problems. And they are getting ready:.
0
0
7
Thank you LinkedIn influencer, but it’s not DeepMind, it’s me
1
0
31
Fun fact: i wanted to name this “XTX Can Be Faster” after trading firm @xtxmarkets . When i was 18yo, Alex Gerko (XTX Markets founder) directly called me by phone. The call should have been a Probability Theory interview. I prepared by solving all questions from all MSU.
We discovered faster way to compute product of matrix by its transpose!. This has profound implications for data analysis, chip design, wireless communication, and LLM training!. paper: The algorithm is based on the following discovery: we can compute
1
2
21
0
2
6
Addendum: . For general matrix product A*B people use Strassen-Winograd algorithm based on 2x2 block-matrix case. They don't use other asymptotically (n -> inf) faster algorithms because of big overhead for small matrix sizes. In fact, the smallest known algorithm with.
2
4
122
Really wonderful work.
Introducing AlphaEvolve: a Gemini-powered coding agent for algorithm discovery. It’s able to:. 🔘 Design faster matrix multiplication algorithms.🔘 Find new solutions to open math problems.🔘 Make data centers, chip design and AI training more efficient across @Google. 🧵
0
2
5
For the last week i tried really hard to squeeze out new knowledge out of o3 - like a new theorem / algorithm / novel combinatorial counterexample. Very little success so far but i highly recommend it as a math/cs literature search tool. It is able to find references and papers.
1
2
12
Am i the only one irritated by gatekeeping from AI companies?. We have o4-mini, which implies existence of o4-full. Yet they don’t talk about this model. Same with unreleased Gemini-2.5-Ultra, Claude 3.7-Opus. I get it they are playing pricing games, but its not a good outlook.
1
1
7
As a winner of international math olympiads who graduated 4 years ago, this cannot be further away from the truth. Top talents every year are way smarter, hard-working, and competitive than before. Don’t listen to these rants by boomers.
This week, a math professor at MIT told me that incoming students are, on average, noticeably worse at math than they used to be. Harvard, of course, just added a remedial math class, Math MA5, "aimed at rectifying a lack of foundational algebra skills among students".
3
1
13
My new favorite tool is prediction markets like Manifold, Polymarket, Kalshi etc. It’s a great reality check tool - if you are consistently wrong in prediction markets, your world model is really wrong and you are in an information bubble. These days so many people live in.
1
2
12
RT @RuoyuSun_UI: If you are interested in GRPO by deepseek, you might want to try ReMax, which is quite related to GRPO but perform better….
0
8
0