bidhan @ NVIDIA GTC

Being at the frontier - by the definition of it - means creating the frontier. You don't get to be at the frontier by following someone else. And creating the frontier often means discoveries that go against the established knowledge. We recently made such a discovery about distributed diffusion model training. A common way to optimize diffusion model training is by ensuring the numerical stability of their generation paths. We found that that's not true for the most efficient distributed diffusion model training architecture. We shared what works instead in our blogpost below. blog.bagel.com/p/stability-qu…

what if told you training coding model to code python is optimizing for a local maxima

the models that were traditionally AR (llms) are becoming diffusion, not the other way around

@philipkiely @baseten i need to get one!

Giving away signed copies of Inference Engineering at NVIDIA GTC — stop by booth 931!

@rokobasili pull up luma.com/nvidia-gtc-gen…

@bidhan We out here

who’s here

Deepseek v4 still not released Alibaba Qwen going closed Western open weights models slacking In these dark times for open source, who will save us? Alliances must be made, brothers must band together! A world of only closed source AI will lead to consolidation of power! Tyranny!

@ClementDelangue @huggingface and soon even frontier models will require 20% of the current cost to train. see @bageldotcom!

They tell you training and running AI model costs billions. That's true for a few frontier labs. But for most real-world use cases? Dramatically lower than you think thanks to open-source. Real examples from @HuggingFace's latest analysis: - Fine-tune a text classification model: <$2k - Train a leading image embedding model: < $7k - Train Deepseek OCR: < $100k - Train a leading machine translation model: <$500k Compare that to GPT-4.5 training (~$300M est.) And the truth is that you don't need a Formula 1 car to pick up groceries. Most tasks are solved just as well by smaller, efficient, targeted models. The mistake everyone makes? Starting with "what's the best AI model?" instead of "what do I need to do?" The future of AI is not just bigger models. It's cheaper, more customized, open models solving specific problems. Explore 100+ real model costs of training and deployment yourself in the study!

In town for NVIDIA GTC? If you're building generative world models or investing in the people who are - we're putting the right people in one room for you tomorrow night in Palo Alto. Co-hosted by Alumni Ventures. Signup link below.

We've spent the last year figuring out how to train frontier diffusion models without a GPU supercluster. Turns out a lot of people training generative models care about the same problem. Come to our event tomorrow to meet them luma.com/nvidia-gtc-gen…

GTC has rave his year. Oh how the times have changed.

all the startups in my network that had good funding and seemed to have a lot of potential couple of years back, but for mysterious reasons didn’t work out are fully remote

we managed to extend DDM to heterogeneous objectives! one step closer to decentralized AI XD tldr: experts trained in complete isolation, with different objectives, no communication - and mixing them beats making them all train the same way. 20-48G memory per expert

@rcadxwn thanks! it's the mission.

@bidhan Killing it! Thanks for open-sourcing it all

Excited to share that Bagel Labs' paper got accepted at CVPR 2026. A lot of the most important diffusion model research has historically stayed inside frontier labs. We're bringing more of that in the open through open science and open infrastructure. In this work we showcase the very counterintuitive advantage of mixing different training objectives (DDPM and Flow-Matching) through an ensemble of diffusion models. This is one of the first ever works to successfully combine diffusion models trained with heterogeneous objectives. See details here: blog.bagel.com/p/heterogeneou…

@astronomerozge1 🥯🥯🥯

@bidhan 🥯

arxiv link arxiv.org/abs/2603.06741

@botir33751732 haha yes. i was just piling on more logics to the argument.

@bidhan Just to clarify I agree with you

with the current ai research neolabs boom, no self respecting researcher should be working for a bank

@botir33751732 why work on moving value created by other people instead of creating value yourself?

@bidhan Intellectual suicide

or hedge fund