Twitxy Saved

Wei Guo

@WeiGuo01

ML PhD, Georgia Tech; Research Intern, NVIDIA; math undergrad, PKU: Generative AI / Diffusion / LLM / MCMC Sampling / Agent-assisted Theoretical Research
Atlanta, GA alexandreguo2001.github.io Joined July 2022
979 Following 473 Followers 160 Posts

Latest posts 5

Wei Guo @WeiGuo01
We have done a major update on distillation and speculative decoding, and release our Nemotron-Labs-Sigma model family on Huggingface. Check it out at huggingface.co/collections/nv…
Kudos to all collaborators and looking forward to the future of continuous diffusion LLMs!
huggingface.co Nemotron-Labs-Sigma - a nvidia Collection Continuous diffusion language models (3B/8B): base, instruct, few-step distilled (NFE 2/4/8/16) and self-speculative variants.
Morteza Mardani Verified account @MardaniMorteza 13h
📣 NVIDIA is releasing a collection of open model weights: Nemotron-Labs-Sigma — 14 checkpoints at 3B/8B.

Language models have stayed discrete. We bring continuous diffusion to language as in vision.

Smooth ODE trajectories (16-D) unlock distillation and steering: 8 TPFs for math, 4 TPFs for code, with accuracy and diversity preserved. 🧵 1/
View quoted post on X
2 replies 10 reposts 31 likes 994 views View on X
Wei Guo @WeiGuo01
Another great work on large-scale continuous diffusion LLMs 👀
Also huge thanks for sharing our work Sigma arxiv.org/abs/2610.02665 in your post! I learned a lot from the insightful discussion with @chandavidlee and @wognsfjq96 during ICLR 2026 on flow language models
arxiv.org Large Language Continuous Diffusion Models Despite the success of discrete diffusion language models (dLMs) for fast parallel decoding, their non-smooth, high-dimensional space hinders trajectory steering for reasoning and inference...
Chanhyuk Lee Verified account @chandavidlee Oct 21
Can flow language models scale to billions of parameters and handle real generative tasks?

Introducing 🌊 Lafa (Large flow language model), a flow LM scaled up to 8B that matches AR quality, with 🚀 2.5× AR throughput at 7B.

Co-led with @ArnavChavan6 @UdbhavBamba.
Joint work with @wognsfjq96 @hongseu33 @AdtRaghunathan @jw9730
View quoted post on X
4 replies 4 reposts 32 likes 1.7K views View on X
Wei Guo @WeiGuo01
Excited to share our latest RL method for diffusion models!
The cord idea behind this simple and effective object: collect final samples from PFODE rollout and their rewards, then compute an advantage-weighted pretraining loss to reinforce/forget good/bad generated samples
Jaemoo Choi @JaemooChoi Oct 9
Simple, efficient, scalable diffusion RL without likelihood ratios?

Meet RVM: Reward-based Velocity Matching

🎯 Rewards directly in velocity space
⚡ Cheap video fine-tuning
🧩 Unifies recent objectives
🎬 Fixes the static-video problem by dynamic motion reward

How it works 🧵
View quoted post on X
0 replies 6 reposts 51 likes 2.8K views View on X
Wei Guo @WeiGuo01
Excited to share my work during summer internship at NVIDIA! We build the first large-scale continuous dLLM, reaching comparable and even better performance compared with masked dLLM trained under aligned budgets, and discover interesting properties its continuous dynamics!
Zhihan Yang Verified account @zhihanyang_ Oct 5
💫 Large Language Continuous Diffusion Models

Presenting Sigma, the first large-scale (3B/8B) continuous diffusion LM (dLM) trained with likelihood.

📈 Comparable with discrete dLMs and AR on math and coding
🎛️ Enables meaningful study of inference-time interventions

Jointly led with Wei Guo (@WeiGuo01), Jean-Marie Lemercier (@jm_lemercier), Simon Welker (@_cobalamin)

Team: @YongganFu @mmkamani7 Sajad Norouzi @julberner @tomasgeffner @karsten_kreis @YongxinChen1 @MoleiTaoMath @jwthickstun @PavloMolchanov Ante Jukic @ArashVahdat @MardaniMorteza
View quoted post on X
6 replies 20 reposts 74 likes 3.5K views View on X
Wei Guo @WeiGuo01
I'll be in Seoul next week for ICML to present the following papers:
DASBS: arxiv.org/abs/2602.08243
DMPO (spotlight): arxiv.org/abs/2510.08233
Rethinking diffusion RL: arxiv.org/abs/2602.04663
MetaDNS: arxiv.org/abs/2605.21722
Please don't hesitate to reach out if you're interested!
arxiv.org Discrete Adjoint Schrödinger Bridge Sampler Learning discrete neural samplers is challenging due to the lack of gradients and combinatorial complexity. While stochastic optimal control (SOC) and Schrödinger bridge (SB) provide principled...
Molei Tao Verified account @MoleiTaoMath Jun 30
Attending #ICML2026? Plz consider visiting us to discuss what we're passionate about -
GenAI (from📚the very theoretical end, all the way to🚀the very applied end), neural sampler, and more.
We'd love to learn from you! Thank you.
View quoted post on X