Timofei Gritsaev (@gritsaev) header
Timofei Gritsaev (@gritsaev) profile picture

PhD at Max Planck Institute for Intelligent Systems | ex-BayesGroup

Germany
Joined July 2020
Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)Photo by Timofei Gritsaev (@gritsaev)
Timofei Gritsaev retweeted
Happy to share our most recent work: An ODE-based Sampler for Discrete Diffusion Models - we introduce a probability-flow analogue for discrete diffusion: deterministic dynamics that preserve its time marginals. - we derive a principled discretization and efficient implementation for both MDM and UDM. - this allows a theoretically grounded consistency distillation objective. arxiv.org/abs/2610.40618 Joint work with @samsongvch, @ericxing, @EricMoulines, @AlainDurmus and Giacomo Zanella We are very excited about this direction and believe that it can be pushed much further! (1/n)
2
41
2
154
6,818
Timofei Gritsaev retweeted
My course Deep Learning 2 is back on Sep 1 🚀 Updated materials, modern DL topics, and guest lectures from researchers & industry experts. Everything is open-source, and the repo just passed 100⭐ github.com/thecrazymage/DL2_…
3
1
8
284
Timofei Gritsaev retweeted
💨 Did you know neural networks can behave like literal ideal gases? 💨 In our #IJCAI2026 paper, we show that at stationarity, scale-invariant NNs surprisingly obey... the ideal gas law 🤯 Thermodynamics helps us to understand how train hyperparams shape final solutions! 🧵👇1/8
1
14
3
48
5,895
🍌🍌🍌 1/ The original GFlowNets paper tried PPO for sampling, but it failed. We figured out why and fixed it. And now PPO beats all objectives on standard problems including molecular graph generation in large spaces.
10
37
1
340
24,809
7/ We then consider molecular graph generation problems: sEH and QM9. These are large problems with over 10^18 and 10^180 states, respectively. Our results are: 1️⃣Ent-PPO achieves the best result! 2️⃣The gap in ELBO between Ent-PPO and value-based methods grows with problem difficulty! Caption: ELBO and EUBO are common metrics measuring inner- and outer-mode coverage, respectively. A higher ELBO and a lower EUBO indicate more accurate sampling.
1
1
6
1,120
8/ So, first, Ent-PPO outperforms value-based approaches, and, second, we assume Ent-PPO shines in more challenging problems! See a detailed Ent-PPO derivation and discussion of important implementation choices in the paper: 📄 Paper: arxiv.org/abs/2606.15793 💻 Code: github.com/tgritsaev/ent-ppo This is a joint work with my amazing coauthors, @akzykova @dtiapkin @nvimorozov!
1
1
1
26
1,172
Timofei Gritsaev retweeted
Why Gaussian diffusion models fail on text data and how to prevent it? ☝️ We find that discrete-like latent spaces are fundamentally bad for continuous diffusions. ☝️ We explain what happens inside, and why self-conditioning and other heuristics improve generation. 🧵👇 1/7
7
35
3
232
23,648
Great work by my dear colleagues!
(1/n) Excited to share our latest work “Learning Shortest Paths with Generative Flow Networks”! We uncover a novel theoretical connection between flow minimization in GFlowNets and finding shortest paths, and develop a learning approach that rivals SOTA in solving Rubik's Cubes!
1
257
Timofei Gritsaev retweeted
Got asked in a review for my ICML paper whether "there are realistic tasks where one needs to sample from a probability distribution given by its unnormalized density" (rephrased for anonymity). Are we cooked?
17
10
6
346
69,146
Timofei Gritsaev retweeted
LOTs of discourse lately about the correctness of the KL-regularization term used in RLVR fine-tuning of LLMs. Which estimator to use? Whether to add it to the reward or loss? What’s even the difference? 🤔 In our new preprint, we evaluate these choices empirically. 🧵 1/n
8
37
9
139
26,466
Timofei Gritsaev retweeted
While frontier labs are announcing their new models, we also want to be part of this parade. So, we’re happy to announce gfnx – a JAX-first library with environments and a single-file baseline implementation for GFlowNet research.
4
10
21
1,463
1/ Can we efficiently learn the destruction process of diffusion samplers? Can we learn not just the drift, but also the variance for all transition kernels? – We answer YES in our recent paper “Adaptive Destruction Processes for Diffusion Samplers” (Oral at NeurIPS 2025 FPI Workshop).
1
9
1
17
2,802
6/ But it’s not limited to small-scale problems. We successfully use these methods to improve outsourced sampling: Bayesian posterior inference in latent spaces of generative models.
1
1
166