Byoungwoo Park

6 posts

Byoungwoo Park

Byoungwoo Park

@bw__park

Graduate Researcher at @kaist_ai Visiting Scholar @GeorgiaTech | Interested in Schrödinger bridge, diffusion models, stochastic optimal control

Atlanta, GA Katılım Mayıs 2024
55 Takip Edilen36 Takipçiler
Byoungwoo Park retweetledi
Guan-Horng Liu
Guan-Horng Liu@guanhorng_liu·
If you're in ICML🇰🇷, come checkout #FAS—a natural (yet w/ nontrivial math 🧐) extension of #Adjoint #Sampling to #functional space for sampling irregular time series such as transition paths. 🗓️ Jul 8 (Wed), 5:00–6:45 PM KST 📍Hall A #3408 Great work by Byoungwoo @bw__park 🫡!
Guan-Horng Liu@guanhorng_liu

📢#Adjoint #Sampling is a new Diffusion Sampler for Boltzmann distribution that - Grounded on stochastic control - Enjoy scalable matching objective - Extremely efficient in energy NFE - Does NOT require/estimate target data Checkout @aaronjhavens talk on Monday #FPI workshop!

English
0
7
44
5K
Byoungwoo Park retweetledi
Jaemoo Choi @ ICML 🇰🇷
Jaemoo Choi @ ICML 🇰🇷@JaemooChoi·
Thrilled to announce that "Rethinking Design space of RL for Diffusion Models" has been accepted at ICML 2026, see you in 🇰🇷!
Jaemoo Choi @ ICML 🇰🇷@JaemooChoi

We proudly present “Rethinking the Design Space of RL for Diffusion Models” showing that ELBO-based likelihood estimation (from the final sample) is the dominant driver of stable, efficient RL fine-tuning. On SD3.5-Medium, we boost GenEval 0.24 → 0.95 in ~90 GPU hours, beating FlowGRPO (4.6×) and DiffusionNFT (2×) efficiency. Great Collab with @YongxinChen1 @YuchenZhu_ZYC @WeiGuo01 @MoleiTaoMath Petr Molodyk, Bo Yuan, Jinbin Bai, Yi Xin 🎥 Video attached 📍Link: arxiv.org/abs/2602.04663 #DiffusionModels #ReinforcementLearning #TextToImage #GenAI

English
2
7
62
6.4K
Byoungwoo Park retweetledi
Wei Guo
Wei Guo@WeiGuo01·
I’ll present two papers at ICLR and I’m happy to chat! (1) Proximal Diffusion Neural Sampler (Apr 23 morning, P3-#411) (2) Complexity Analysis of Normalizing Constant Estimation: from Jarzynski Equality to Annealed Importance Sampling and beyond (Apr 23 afternoon, P4-#4509)
Wei Guo@WeiGuo01

How annealing helps overcoming multimodality? In our ICLR 2025 paper openreview.net/forum?id=P6IVI… and preprint arxiv.org/abs/2502.04575, we established the first complexity bound for annealed sampling and normalizing constant (⇔free energy) estimation under weak assumptions on target!

English
0
8
36
4K
Byoungwoo Park retweetledi
Wei Guo
Wei Guo@WeiGuo01·
How to bring the speed & precision of continuous adjoint matching (AM) to discrete neural samplers? Introducing discrete adjoint Schrödinger bridge sampler (DASBS): a unified framework for authentic discrete AM! 🎲✨ Joint work with @JaemooChoi et al.: 📄arxiv.org/abs/2602.08243
Wei Guo tweet media
English
2
10
46
8.8K
Byoungwoo Park retweetledi
Jaemoo Choi @ ICML 🇰🇷
Jaemoo Choi @ ICML 🇰🇷@JaemooChoi·
We proudly present “Rethinking the Design Space of RL for Diffusion Models” showing that ELBO-based likelihood estimation (from the final sample) is the dominant driver of stable, efficient RL fine-tuning. On SD3.5-Medium, we boost GenEval 0.24 → 0.95 in ~90 GPU hours, beating FlowGRPO (4.6×) and DiffusionNFT (2×) efficiency. Great Collab with @YongxinChen1 @YuchenZhu_ZYC @WeiGuo01 @MoleiTaoMath Petr Molodyk, Bo Yuan, Jinbin Bai, Yi Xin 🎥 Video attached 📍Link: arxiv.org/abs/2602.04663 #DiffusionModels #ReinforcementLearning #TextToImage #GenAI
English
1
28
178
33.2K
Byoungwoo Park retweetledi
Guan-Horng Liu
Guan-Horng Liu@guanhorng_liu·
Adjoint Matching works great for fine-tuning diffusion models with reward gradients. How about #AM for #diffusionLLMs with #nondifferentiable #rewards? Does "discrete adjoint" even exist ... and how? 🤔 📢 Introduce #DiscreteAdjointMatching (#DAM)—a unifying AM for discrete generative models, accepted to #ICLR2026 🇧🇷 Work done with my amazing intern @oswinso and @RickyTQChen, Brian, Chuchu 🙌 📰 arxiv.org/abs/2602.07132
GIF
English
1
11
94
8.2K

Keşfet

@JaemooChoi @YongxinChen1 @YuchenZhu_ZYC @WeiGuo01 @MoleiTaoMath @oswinso @RickyTQChen @elonmusk