CEO @SophontAI | Founder @MedARC_AI | PhD at 19 (2023) | ex Research Director Stability AI | Biomed. engineer @ 14 | TEDx talk➡bit.ly/3tpAuan
- Tweets 21,750
- Following 1,555
- Followers 91,395
- Likes 133,886
Pinned Tweet
I will be in SF/Bay Area for the next couple weeks for:
- SF Tech Week
- COLM
- Open-source AI week
Also will be speaking at an event (link in reply)
I would love to catch up with friends, colleagues, and collaborators! If you wanna hang out, DM me!
5
1
37
3,326
I will be in SF/Bay Area for the next couple weeks for:
- SF Tech Week
- COLM
- Open-source AI week
Also will be speaking at an event (link in reply)
I would love to catch up with friends, colleagues, and collaborators! If you wanna hang out, DM me!
5
1
37
3,326
Event: From PhD to Founder -#SFTechWeek
Wednesday, Oct 7, 6:00pm – 8:30pm
Partiful: partiful.com/e/QqnqXMeVkv5Kv…
3
642
It's remarkable that the reviewer of the first scientific paper I wrote became a long-time Twitter mutual!
Thank you for your support Dr. Sandre!
Indeed I have had an awesome journey in science over the past 10 years :)
I remember Tanishq when I reviewed the 1st research paper you wrote... What a journey accomplished since then! ;-)
2
2
10
2,315
Can't believe it was A DECADE AGO when I started my journey at UC Davis at 13 yrs old!
I went on to graduate with a bachelor's (2018) and PhD (2023) in biomedical engineering.
I was just a young kid when I started! I've grown so much since then :)
#firstdayofschool2016 in @ucdavis -reading campus map to get to classes & dodging bikes- bike population seems to be more than student pop!
18
4
151
12,523
it's cool that one of the biggest youtubers is playing around with AI training and sharing it with millions of people!
but fwiw this is like a year old in terms of doing uncensored finetuning
he should catch up to the frontier and start a PewDieAI open-source neolab 🤣
PewDiePie unveiled Ajax, his own "uncensored" AI model built to run on home PCs, and said OpenAI banned him twice while he was making it
One ban was for "distillation," using another AI's outputs to train his own. "How did they even know?" he said
6
29
4,864
hmmm what's this? 👀
I asked people which neolabs under $10B have the highest talent density:
32 votes: Core Automation (@MillionInt, @_arohan_)
26 votes: Periodic Labs (@LiamFedus, @ekindogus)
20 votes: Flapping Airplanes (@spectorb, @amspector100, @aidanmantine)
18 votes: Standard Intelligence (@G413N, @devanshpandey)
8 votes: Ricursive (@annadgoldie, @Azaliamirh)
5 votes each:
- Ineffable (David Silver)
- Mirendil (@bneyshabur, @HarshMeh1a, @shayan_, @tararezaeikh)
- Recursive (@RichardSocher, @_rockt, @timshi_ai, @josh_tobin_, @CaimingXiong, @jeffclune, @tydsh, Alexey Dosovitskiy)
3 votes each:
- Applied Compute (@ypatil125, @rhythmrg, @lindensli)
- Isara (@ezhang7423, @hegasz)
- Prime Intellect (@vincentweisser, @johannes_hage)
2 votes each:
- Elorian (@AndrewDai, @yinfeiy, @SethInternet)
- Goodfire (@eric_ho, @DanJBalsam, @banburismus_)
- Inherent (@tantumscollins, @edwardfhughes, @LouisKirschAI, @kallyaleksiev)
- Trajectory (@rronak_, @QuantumArjun, @MichaelElabd)
- World Labs (@drfeifei, @jcjohnss, @BenMildenhall, @chlassner)
People couldn't pick a company they work at or founded. Disclosure: I'm a small investor in Applied Compute, Factory, Standard Intelligence, Trajectory and Wafer. I didn't vote.
Continued:
4
1
38
10,741
Come check out our MedARC journal club presentations, next one is on Monday about training foundation models for intracranial EEG:
Continuing our Journal Club series on Monday October 5th at 11am PT!
Ben Tang (researcher at Duke) will be presenting his paper "Pretraining for Sample-Efficient Neural Interfaces"
To attend, join the Discord and check our website!
1
2
11
2,863
Tanishq Mathew Abraham, Ph.D. retweeted
Continuing our Journal Club series on Monday October 5th at 11am PT!
Ben Tang (researcher at Duke) will be presenting his paper "Pretraining for Sample-Efficient Neural Interfaces"
To attend, join the Discord and check our website!
1
2
3
3,327
this looks awesome!
great to see more open initiatives to study the development of frontier capabilities, this is significantly lacking right now...
Today we're unveiling Trillium Labs @trillium_labs, a new non-profit to foster the open science of frontier AI. We're building open post-training recipes and will expand into open infra to study RSI, reward-hacking, multi-agent systems, and whatever comes next.
We're built around the theory of change that you need more eyes to solve hard technical problems. We have faith in the scientific methods and communities that humanity has built, and worry that AI is becoming too closed to utilize them.
Trilliums are wildflowers that bloom briefly in the spring, before the forest canopies fill out. Though they are small, they lay the foundation for the cycles of growth and nourishment through the rest of the year. At Trillium Labs, the recipes will be the slow nutrients for the seasons and the model releases will be the blooms. Building an institution dedicated to this is needed because, much as nature’s trilliums are slow to expand and grow, the open-ecosystem needs time and dedicated resources to catch up.
I co-founded with with a long-time friend and collaborator Tom Zick (@thesezickbeats). We're hiring (full time + student collabs/interns), we're fundraising, and we're looking for compute. Please get in touch if you're interested in helping out. Offices based in the Bay Area and Cambridge MA, remote okay.
I’m in the Bay Area until for The Curve and COLM to connect with people who are interested. We’re thankful to have initial support from Halcyon Futures and Schmidt Sciences with more funding en route to enable our ambitions of scaling. Our advisors @Thom_Wolf, @HannaHajishirzi, @gneubig and @ctnzr have been instrumental to building the ecosystem that exists today, and I’m stoked to get to keep working with them.
1
33
4,989
Sharpening Tax in Post-Training
"Our surprising finding is that pre-trained LLMs, equipped with a light inference harness, can serve as capable agents. Despite far lower accuracy (pass@1), they often surpass their post-trained counterparts in solution coverage (pass@K) given a sufficient test-time budget."
"we propose Sharpening Tax, a diagnostic metric that quantifies the loss in test-time scalability after post-training"
"we present posterior-tempered group sampling (PTGS), a simple plug-and-play Bayesian sampler that adapts the sampling temperature per prompt to its estimated difficulty."
11
17
123
8,062
project page: changdaeoh.github.io/sharpen…
code: github.com/changdaeoh/sharpe…
paper: arxiv.org/abs/2610.01509
5
841
Trust the Critic More
"We introduce Actor-Critic with Action Chunking (AC2) that removes the need to roll every trajectory to completion. AC2 instead assigns credit to action chunks: short continuations of prefixes of past trajectories. A learned critic scores the state reached at the end of each action chunk, allowing the policy to update without observing a terminal reward."
"First, we introduce local readiness which uses critic-based updates on a problem only when the critic is sufficiently accurate on that particular problem. Second, when available, we provide the critic with a reference solution from a previous successful rollout. Third, we assign credit over action chunks of 10k tokens rather than individual tokens, giving the critic a more meaningful portion of the trajectory to evaluate."
Link: arxiv.org/abs/2609.39247
6
25
160
10,202
How Much Is an AI Token Worth? Scaling Laws for Wild AI-Generated Web Text
"After applying FineWeb quality filtering, we find that 27.5% of tokens from June 2026 web data are labeled as AI-generated by Pangram, rising to 31.1% by August."
"How does AI text in the wild affect language model pretraining? To answer this question, we pretrain 800 language models, varying the ratio of added AI tokens to human tokens, and fit scaling laws to held-out losses on both human and AI-generated text. For data-starved models, adding AI tokens to pretraining data initially lowers loss on human text, but the benefit saturates as more are added and quickly *reverses* into harm. For models trained on high budgets of human text, AI tokens raise loss almost immediately, whil"e the same number of fresh human tokens keeps lowering it."
"We propose a new scaling law with separate benefit and harm terms that allows the value of an AI token to change sign while also reducing to Chinchilla in the absence of AI text. When fit on smaller models, our scaling law predicts the effect of AI text on held-out human-text loss for models up to 3.6x larger with 41% lower error than the best existing law over all AI ratios."
link: arxiv.org/abs/2609.40295
4
3
30
3,037
Frontier VLMs perception of biomedical images still struggle significantly, and is one of the primary bottlenecks in frontier medical capabilities.
The MMBU challenge aims to address this, and it is officially about to start!
3 tracks. $100K+ in compute and prizes. Oct 1–Dec 31.
Highly recommend checking it out, last chance to sign up!
Register: luma.com/28k1tyd3
Link: akiranishii.github.io/mmbu-c…
6
9
78
5,257
Great evals, very promising! Looking forward to trying it out...
It's odd to me Google DeepMind never reports any medical capabilities evals. GDM has quite a strong medical AI team and publish really interesting papers in the space, but none of it seems to be talked about for Gemini.
Imo, Google should be paying a lot more attention to this when so many people rely on their model.
Introducing Gemini 4 Argon – our new frontier model.
It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
10
85
7,094