Nikhil Barhate @nikhilbarhate99
ML @scale_AI | prev @AMD @mila_quebec nikhilbarhate99.github.io San Francisco, CA Joined June 2015-
Tweets1K
-
Followers203
-
Following813
-
Likes3K
We have been doing work on scaling laws for off-policy RL for some time now and we just put a new paper out: arxiv.org/abs/2508.14881 Here, @preston_fu @_oleh lead a study on how to best allocate compute for training value functions in deep RL: 🧵⬇️
Instructions/reasoning are now everywhere in retrieval - we want embeddings to do it all! 🚀 But... is it even possible? 🤔 Turns out, it's not possible for single-vector models 😱 theoretically and empirically! To make it obvious we OSS a simple eval SoTA models flop on! 🧵
@suchenzang @yaroslavvb No way!! I literally wrote libheatmap precisely to make heatmaps of event locations in dota2, in 2013 lucasb.eyer.be/articles/color…
very recently (as of v0.1.3) figured out what i think is the "right" way to handle Rubric-level state + objects being made available to reward functions inside verifiers previously, you'd just declare extra things globally (def an anti-pattern, always bugged me) and i'd manually…
VLAs offer an avenue for generalist robot policies; however, naively following the action predictions leads to brittle or unsafe behaviours. We introduce VLAPS, which integrates model-based search with pre-trained VLA policies to improve performance without additional training.
Cooking up cool stuff at work 🍜🤖 had a great time building model debate for data quality!
Cooking up cool stuff at work 🍜🤖 had a great time building model debate for data quality!
I've written the full story of Attention Sinks — a technical deep-dive into how the mechanism was developed and how our research ended up being used in OpenAI's new OSS models. For those interested in the details: hanlab.mit.edu/blog/streaming…
I spent the past months investigating: Can we trust reasoning models' CoTs? Researchers showed that LLMs aren't always faithful, but that's not the full story. LLMs are very faithful when the reasoning is complex, and unfaithful CoTs remain monitorable! Check out my latest work🥳
I spent the past months investigating: Can we trust reasoning models' CoTs? Researchers showed that LLMs aren't always faithful, but that's not the full story. LLMs are very faithful when the reasoning is complex, and unfaithful CoTs remain monitorable! Check out my latest work🥳
Failing on 𝐥𝐚𝐫𝐠𝐞-𝐬𝐜𝐚𝐥𝐞 𝐑𝐋 with VeRL? ⚠️ Mixing inference backend (𝐯𝐋𝐋𝐌/𝐒𝐆𝐋𝐚𝐧𝐠) with training backends (𝐅𝐒𝐃𝐏/𝐌𝐞𝐠𝐚𝐭𝐫𝐨𝐧) 𝐬𝐞𝐜𝐫𝐞𝐭𝐥𝐲 𝐭𝐮𝐫𝐧𝐬 𝐲𝐨𝐮𝐫 𝐑𝐋 𝐢𝐧𝐭𝐨 𝐨𝐟𝐟-𝐩𝐨𝐥𝐢𝐜𝐲 — even if they share the same weights! 📉 Blog:…
Genie 3 is here - it can generate an entire world simulation that you can interact with in real-time, just from a text prompt! It's pretty mind-blowing really when you stop to think about it, and it's rapidly improving - one day we will be able to build the Holodeck for real!
🚀 Excited to share our #CoRL2025 paper! See you in Korea 🇰🇷!🎉 We present ParticleFormer, a Transformer-based 3D world model that learns from point cloud perception and captures complex dynamics across multiple objects and material types ! 🌐 Project website:…
July has been a big month for Viser! - Released v1.0.0😊 - We did some writing Some demos👇
Excited to have our AI research published in @Nature today. Proud of the @ProfluentBio team and the extensive final version available under open-access. OpenCRISPR is a milestone. It's the first successful demonstration of editing the human genome with a molecule fully designed…
Excited to have our AI research published in @Nature today. Proud of the @ProfluentBio team and the extensive final version available under open-access. OpenCRISPR is a milestone. It's the first successful demonstration of editing the human genome with a molecule fully designed…
Excited to share Flow Matching Policy Gradients: expressive RL policies trained from rewards using flow matching. It’s an easy, drop-in replacement for Gaussian PPO on control tasks.
Fine-tuning pre-trained robotic models with online RL requires a way to train RL with expressive policies Can we design an effective method for this? We propose EXPO, a sample-efficient online RL algorithm that enables stable fine-tuning of expressive policy classes (1/6)
There is no fucking way i wasnt aware of this work that came out this may, literally DAVID SILVER and JEFF DEAN is coauthor ???
Engineers spend 70% of their time understanding code, not writing it. That’s why we built Asimov at @reflection_ai. The best-in-class code research agent, built for teams and organizations.
Diffusion/flow policies 🤖 sample a “trajectory of trajectories” — a diffusion/flow trajectory of action trajectories. Seems wasteful? Presenting Streaming Flow Policy that simplifies and speeds up diffusion/flow policies by treating action trajectories as flow trajectories! 🌐…
Tokenization has been the final barrier to truly end-to-end language models. We developed the H-Net: a hierarchical network that replaces tokenization with a dynamic chunking process directly inside the model, automatically discovering and operating over meaningful units of data
Introducing Hierarchical Surgical Robot Transformer (SRT-H), a language-guided policy for autonomous surgery🤖🏥 On the da Vinci robot, we perform a real surgical procedure on animal tissue. Collaboration b/w @JohnsHopkins & @Stanford
Miles Grimshaw @milesgrimshaw
12K Followers 4K Following Thrive Capital. @cursor_ai @chaidiscovery @doji_com @langchainai @benchling @monzo @latticehq @segment @airtable
Jyoti Mann @jyoti_mann1
3K Followers 4K Following Tech Reporter @businessinsider prev @FT + hedgie consultant. 📧: [email protected] (my views)
Brian Zhan @brianzhan1
3K Followers 2K Following Investing in early stage AI @CRV. Seed/A: @Reflection_AI, @SkildAI, @DynaRobotics, @LanceDB, Lepton (acq NVIDIA), @VoyageAI (acq MongoDB), @SDFLabs (acq dbt)
Yklielaup @Yklielaup51734
21 Followers 959 Following
Nuvalp @Nuvalp13493
1 Followers 424 Following
AI Agent @AAgent67742
1 Followers 103 Following
shayan @shayan4m1r
1 Followers 184 Following
Aniket Pallav @aniketpallav
1K Followers 4K Following 'Tech for good' believer, PgM In free time:aviation enthusiast, start-ups, personal finance, climate change, foodie
Shefali @shefalisastry
373 Followers 569 Following data & rocket scientist. anti business business student @harvard, intern @NFX. ex @spacex, @nextdoor
Vikram Mishra @hypercosmac
1K Followers 728 Following Design x Hardware x Autonomy. prev @thezaptrack
Wormu @Wormu335
9 Followers 190 Following
PureLoveEvangelist(sm... @hamzaiandafirst
474 Followers 3K Following Sin was necessary; It's always in the moments just before death; top 10 functional health experts in the world epistemic superconnector
Noah Jacobson @noahajake
43 Followers 326 Following ML Research at ScaleAI. Formerly at Amazon, Stanford
Chenchen Ye @chenchenye_ccye
832 Followers 914 Following CS PhD student @UCLA, Intern @scale_AI | Prev Intern @MSFTResearch | Prev Undergrad @NUSingapore | Generative Models
Ameen Patel @Ameen_ml
1K Followers 1K Following Inference @PrimeIntellect, prev @togethercompute, @AmazonScience, @uwaterloo
Recombinate Health @People1_team
939 Followers 5K Following I’m Johannon Olson from Recombinate Health. We give care teams superpowers with our Care-Team-Centric-Growth process.
Haoyu Xiong @Haoyu_Xiong_
3K Followers 2K Following PhD student @MIT_CSAIL | Prev @Stanford @CMU_Robotics #Robot_Learning
Edarsaw @Edarsaw6127
11 Followers 709 Following
Melvin Wilderman @MelvinWild93883
44 Followers 3K Following
Alexander Koch @alexkoch_ai
6K Followers 265 Following Founder & CEO at Tau Robotics (@taurobots) | Z Fellow | Emergent Ventures Fellow
Tu Trinh @thetututrain
37 Followers 124 Following Aka Alina Trinh. ML research engineer @scale_AI | EECS MS @UCBerkeley @CHAI_Berkeley @berkeley_ai
Jamison Wintheiser @JamisonWin48058
23 Followers 2K Following
Kent Williams @_kentw
1K Followers 1K Following building in consumer at @retrodotapp ⁂ prev @stripe, co-founder @bitcafe_app
sarv @SarvasvKulpati
10K Followers 2K Following Making computers fun again https://t.co/cUc86o7fBr CS+Cogsci @UCBerkeley YT: https://t.co/OR3L2OZJ8A
Akira Yoshiyama ⁂ @yoshiyama_akira
2K Followers 2K Following research @ETH_AI_Center @Tufalabs | comp eng @UWaterloo | third-space building @socraticainfo
Linda Tong @lktong_
995 Followers 729 Following gonna ditch the obligatory alphabet soup, even if it’s just for a little while https://t.co/zts8mzuAC1
Augustine Mavor-Parke... @MavorParker
357 Followers 419 Following RL environments as co-founder of @VmaxAI at @southpkcommons, prev: RL PhD at @ai_ucl, @redwood_ai, @CSHL, @illumina
Trythare @TrytharebOt0
84 Followers 2K Following
~/mehul/ @luhemarora
168 Followers 149 Following frolicking in the garden of technology @southpkcommons @lightbulbml (acq.) @stanford
Roop Pal @roop_pal_
160 Followers 63 Following making building affordable by using ai to read blueprints @bild_ai. ex-google, ex-waymo, 1x exited founder. graduated columbia cs at 19.
Dorte @Dorte2388
16 Followers 1K Following
daft @strahead
1 Followers 251 Following
Scortewr @Scortewrq11GLH
65 Followers 1K Following
Slnoytr @SlnoytrUMb_al
36 Followers 2K Following
Tyler Griggs @tyler_griggs_
553 Followers 349 Following CS PhD student @UCBerkeley Sky Lab, co-leading @NovaSkyAI and building SkyRL | Previously @GoogleCloud infra | @Harvard 2020
Michael Galkin @michael_galkin
7K Followers 330 Following Senior Research Scientist @GoogleAI. Prev: @Intel, Postdoc @Mila_Quebec & McGill. Graph Learning & LLMs. Grandmaster of 80's music (according to Spotify)
Commonwealth Fusion S... @CFS_energy
16K Followers 327 Following We’re on a mission to deliver clean fusion energy to the planet fast enough to matter for humanity’s biggest challenges.
lele @CherrilynnZ
2K Followers 885 Following in meandering pursuit of aesthetics and function product designer / formerly psych @cal / @writewithprl
UFB - Ultimate Fighti... @UFBots
116K Followers 19 Following Major Humanoid Fighting League | Remote Control from Anywhere | Built by @frodobots
Fang @FangSystems
5K Followers 310 Following Co-Founder + Founding Robotics Engineer @ Paddington Robotics || @join_EF S24 || Aerospace and robotics engineering 🚀🤖⚙️ Projects in highlights tab!
Madison @Madisonkanna
74K Followers 338 Following learning out loud. eng @basetenco. I use vim btw. https://t.co/YaAPIDcMhX
Shengjia Zhao @shengjia_zhao
52K Followers 230 Following Chief Scientist @ Meta MSL. Formerly MTS @ OpenAI, PhD @ Stanford. I train models. All opinions my own.
jane zhang @jjanezhang
2K Followers 777 Following living life 🎉 | agents & llm training @dbrxmosaicai @dukeu I write monthly :)
Khurram Javed @KhurramJaved_96
2K Followers 151 Following Developing algorithms for real-time reinforcement learning on robots. Research Scientist at Keen, a startup led by John Carmack. Prev ~ PhD with Richard Sutton
Haoyu Xiong @Haoyu_Xiong_
3K Followers 2K Following PhD student @MIT_CSAIL | Prev @Stanford @CMU_Robotics #Robot_Learning
Kevin Lu @_kevinlu
9K Followers 215 Following @thinkymachines. formerly: - @openai: RL, synthetic data, efficient models - @berkeley_ai: decision transformer, universal computation
Trapit Bansal @TrapitBansal
32K Followers 247 Following AI Research @Meta | Co-Creator of OpenAI o1 | Previously @OpenAI, @MSFTResearch, @GoogleAI, @facebook, @iiscbangalore, and undergrad @IITKanpur
Shefali @shefalisastry
373 Followers 569 Following data & rocket scientist. anti business business student @harvard, intern @NFX. ex @spacex, @nextdoor
Karl (in SF) @kaarelkaarelson
124 Followers 273 Following AI @ Dartmouth College | Building payment layer for AI agents
Vikram Mishra @hypercosmac
1K Followers 728 Following Design x Hardware x Autonomy. prev @thezaptrack
Noah Jacobson @noahajake
43 Followers 326 Following ML Research at ScaleAI. Formerly at Amazon, Stanford
Chenchen Ye @chenchenye_ccye
832 Followers 914 Following CS PhD student @UCLA, Intern @scale_AI | Prev Intern @MSFTResearch | Prev Undergrad @NUSingapore | Generative Models
Ameen Patel @Ameen_ml
1K Followers 1K Following Inference @PrimeIntellect, prev @togethercompute, @AmazonScience, @uwaterloo
Letterboxd @letterboxd
739K Followers 2K Following 👀🎥🎬 A global community of film lovers. Share your taste in movies via our free apps for iOS, Android, Apple TV, and on the web.
Impressions @impression_ists
164K Followers 2K Following I would rather die of passion than of boredom - V. van Gogh.
Reborn @reborn_agi
8K Followers 392 Following The open ecosystem for AGI robots. https://t.co/QpUF4jsec3
Tu Trinh @thetututrain
37 Followers 124 Following Aka Alina Trinh. ML research engineer @scale_AI | EECS MS @UCBerkeley @CHAI_Berkeley @berkeley_ai
Eric Zhang @ekzhang1
15K Followers 460 Following Computer systems person, interaction designer. founding eng @modal → dreams of: a simpler, more honest, more human sort of software (people are good, be kind!)
ella schlaghecke @ella_schlags
1K Followers 853 Following i run a policy hackathon and i love people 💃🏻 🍓🌟 @scholarsHQ @cansbridgeproj
sarv @SarvasvKulpati
10K Followers 2K Following Making computers fun again https://t.co/cUc86o7fBr CS+Cogsci @UCBerkeley YT: https://t.co/OR3L2OZJ8A
Jay @jayendra_ram
2K Followers 898 Following building simulations. founder @hud_evals, prev cs+physics @columbia, @ycombinator
Kent Williams @_kentw
1K Followers 1K Following building in consumer at @retrodotapp ⁂ prev @stripe, co-founder @bitcafe_app
Linda Tong @lktong_
995 Followers 729 Following gonna ditch the obligatory alphabet soup, even if it’s just for a little while https://t.co/zts8mzuAC1
Tesla Optimus @Tesla_Optimus
562K Followers 11 Following A general purpose, bi-pedal, humanoid robot capable of performing tasks that are unsafe, repetitive or boring.
CASΞY @caseykcaruso
33K Followers 3K Following Managing Partner @topology_vc | Prev: engineer @Google → EIR @bessemerVP → for a hot sec @harvard → investment partner @Paradigm
Joanne Peng @JoanneZPeng
3K Followers 579 Following aging research, molecular tool dev, neuroai | https://t.co/eEL2wWsnt8 @mit @princeton @thielfellowship | 🇨🇦
おしるこ🥗 @oshiruko_s2
406K Followers 2K Following 物語を感じる絵を描きたいです。i/h @ows28888888 🙏Don’t use my artworks without permission. 🙅♀️I don't accept commissions.
Humanoid Botangelist @GoingBallistic5
21K Followers 598 Following Diary of quotidian musings about the humanoid botanical garden
Augustine Mavor-Parke... @MavorParker
357 Followers 419 Following RL environments as co-founder of @VmaxAI at @southpkcommons, prev: RL PhD at @ai_ucl, @redwood_ai, @CSHL, @illumina
Igor Kulakov @ihorbeaver
7K Followers 439 Following Building a general-purpose robot that is more effective than a humanoid. Cofounder of @viktor_vrp. Backed by @fdotinc. Seed soon.
~/mehul/ @luhemarora
168 Followers 149 Following frolicking in the garden of technology @southpkcommons @lightbulbml (acq.) @stanford
Zelda @zeldapoem
9K Followers 1K Following Founder/ chief magician @nautilusquest – a fully funded residency in SF for young polymaths. @1517fund believed in me first
Pollen Robotics @pollenrobotics
6K Followers 213 Following French Robotics company Doing open source robots made by open minded humans to explore real-world applications Join our community : https://t.co/8j2MrPPHzGTrends for United States
You might like
