Subham Sahoo

This account is verified. Learn more

@ssahoo_

Pioneering Diffusion LLMs | Team Lead @mbzuai - IFM | PhD @cornell

San Francisco, CA s-sahoo.com Joined June 2010

93 Following 4.1K Followers 1.1K Posts

Latest posts 5

Pinned post

Subham Sahoo @ssahoo_ Oct 6, 2025

🎓 Officially a doctor now 😊!!!

As a first-gen college kid, this moment means the world to me.

Grateful beyond words to all my mentors who’ve guided me along the way — from @GMartius who first introduced me to research back in 2017, to @volokuleshov who sparked my love for generative modeling, and finally to @jwthickstun and @Jimantha for their incredible mentorship through the final stretch of my PhD. ❤️

Subham Sahoo @ssahoo_ 18h

Ricky’s OG Neural ODEs paper made me fall in love with generative modeling

Hadi M. Dolatabadi Verified account @hmdolatabadi 1d

Recently, I've been missing the papers that made me fall in love with the field during my PhD. So I've decided to revisit a few of them in a "paper nostalgia" series of mine.

#3: "Augmented Neural ODEs" by Dupont et al from NeurIPS 2019. I love how much this paper explains with a few simple pictures. It finds a basic limit of neural ODEs, and the fix is just as simple as giving the dynamics extra dimensions.

1/20

View quoted post on X

Subham Sahoo @ssahoo_ 1d

Great blog post by @JonathanGeuter and @KempnerInst on our recent Uno paper.

Autoregressive LLMs + Diffusion adapters >>> Training Diffusion LLMs from scratch or Speculative Decoding with small drafters

Stay tuned for a follow-up to the Uno paper with comparisons against DFlash2 and DSpark.

kempnerinstitute.harvard.edu/research/deepe…

Subham Sahoo Verified account @ssahoo_ Sep 4

Diffusion LLMs have two limitations relative to AR models:
(1) Lower quality, and
(2) Slower inference at large batch sizes.

We address this "Uno"
> Retains the AR architecture of LLMs
> Each layer has two sets of weights: AR weights and Diffusion weights
> Diffusion weights enable parallel sampling from the AR distribution losslessly

Results:
💥Faster than all speculative decoding methods: DFlash and EAGLE-3
🔥 Beats ALL diffusion LLMs: Mercury 2, Diffusion Gemma, Llada

Links to the paper, models, code below

View quoted post on X

Subham Sahoo @ssahoo_ 2d

You might as well short all tech stocks.

Nvidia goes down, it takes everybody down with it

Culper Verified account @CulperResearch May 13

NEW: We are short Nvidia $NVDA. We believe Nvidia has a significant China problem. Please see our full report and disclosures now available on our website, link in bio.

View quoted post on X

Subham Sahoo @ssahoo_ 5d

Discrete LLM Tutorial at AACL 2026 🔥

Daisuke OBA Verified account @dai0NLP 5d

I’ll be giving a tutorial on diffusion language models at AACL-IJCNLP 2026 on Nov. 10
w/ @Bollegala
🔗 dllm-aacl.github.io
(Slides coming soon!)
#aacl2026 @aaclmeeting

💭Grateful for the opportunity to organize my perspective on this fast-moving field and share it with the NLP community. This will be my first tutorial, so it’s a bit of a personal challenge for me as well. Looking forward to the discussions :)
View quoted post on X