Abhay Gupta @gupta__abhay
Scaling and efficiency lead @DbrxMosaicAI | Previously @CerebrasSystems @CMU_Robotics | Making GPUs and agents go brrrr !! San Francisco, CA Joined November 2014-
Tweets1K
-
Followers390
-
Following2K
-
Likes4K
What if you could reliably monitor, evaluate, and control your AI’s behavior with a single, adaptable tool – no deep expertise required? Databricks’ new Prompt-Guided Reward Model brings together reward modeling and judging to do just that. PGRM is your AI’s quality control…
This thread captures what’s at the heart of empiricism and converting futuristic ideas to commodities everyone can benefit from!!! Scaling good science and infra is the only path to the AI-integrated future we want for ourselves.
This thread captures what’s at the heart of empiricism and converting futuristic ideas to commodities everyone can benefit from!!! Scaling good science and infra is the only path to the AI-integrated future we want for ourselves.
There’s probably better sushi is Los Altos! Also an under-explored area for sure.
There’s probably better sushi is Los Altos! Also an under-explored area for sure.
Agents are the future! We’re adding an amazing team to build it with us.
Agents are the future! We’re adding an amazing team to build it with us.
In all respects, numer still go up !!!
This!!
Bro just out here spitting facts ngl !!
Not that I have a favorite recent project, but... 🧵 LLM judges are the popular way to evaluate generative models. But they have drawbacks. They're: * Generative, so slow and expensive. * Nondeterministic. * Uncalibrated. They don't know how uncertain they are. Meet PGRM!
Not that I have a favorite recent project, but... 🧵 LLM judges are the popular way to evaluate generative models. But they have drawbacks. They're: * Generative, so slow and expensive. * Nondeterministic. * Uncalibrated. They don't know how uncertain they are. Meet PGRM!
Ever wonder what it'd look like if an LLM Judge and a Reward Model had a baby? So did we, which is why we created PGRM -- the Prompt-Guided Reward Model. TLDR: You get the instructability of an LLM judge + the calibration of an RM in a single speedy package (1/n)
Really excited about ALHF, new work from our research team that lets users give natural language feedback to agents and optimizes them for it. It sort of upends the traditional supervision paradigm where you get a scalar reward, and it makes AI more customizable for non-experts.
Since joining @databricks, our research team has been hard at work on Agent Bricks, a new product that helps enterprises develop state-of-the-art domain-specific agents. We are now releasing a research blog about Agent Learning from Human Feedback (ALHF) databricks.com/blog/agent-lea…
In 2016, @sama and I first met. @OpenAI was a vision. @CerebrasSystems was powerpoint. Sam and the OpenAi founders became one of the early investors in @CerebrasSystems. In the following years, the Cerebras and OpenAI frequently met to explore working together. But the timing…
RLVR and test-time compute are a powerful combo for enterprises, so much so that @databricks now leads overall BIRD single-model leaderboard. This isn't about BIRD, though. It's an example of what our customers are accomplishing in their domains with our RL recipe in Agent Bricks
RLVR and test-time compute are a powerful combo for enterprises, so much so that @databricks now leads overall BIRD single-model leaderboard. This isn't about BIRD, though. It's an example of what our customers are accomplishing in their domains with our RL recipe in Agent Bricks https://t.co/Ut7l0Xor2M
This is just a glimpse of what our RL stack can do. We’re only getting better by the day @DbrxMosaicAI !!
This is just a glimpse of what our RL stack can do. We’re only getting better by the day @DbrxMosaicAI !!
Ever since the release of Cerebras inferencing in HotChips2024, Cerebras has been handing Groq massive Ls. DeepSeek R1 Llama3 70B: Cerebras: 2256 tok/s/user Groq: 398 tok/s/user
Ever since the release of Cerebras inferencing in HotChips2024, Cerebras has been handing Groq massive Ls. DeepSeek R1 Llama3 70B: Cerebras: 2256 tok/s/user Groq: 398 tok/s/user https://t.co/6jWnB0dI86
I'm at ICML 🇨🇦 and I'm hiring at @databricks. Visit our booth if you're interested. My scientific focus: It's 1972 in AI, there's an AI crisis, Dijkstra isn't here to save us, and maybe RL can. Why Databricks? The long road to AGI is being paved here and we have the real evals 🧵
Why you should stop working on RL research and instead work on product // The technology that unlocked the big scaling shift in AI is the internet, not transformers I think it's well known that data is the most important thing in AI, and also that researchers choose not to work…
Got a chance to measure Maximum Achievable Matmul TFLOPS on NVIDIA B200. With each new NVIDIA generation the efficiency keeps on dropping: A100: 86.9% H100: 80.3% B200: 77.6% The updated table is here: github.com/stas00/ml-engi…
Excited to finally talk about our findings! The awesome @saanarkethayan will also be at #icml2025 if you want to dive deeper on any aspect of the paper.
Excited to finally talk about our findings! The awesome @saanarkethayan will also be at #icml2025 if you want to dive deeper on any aspect of the paper.
Paper is here: arxiv.org/abs/2502.05967 If you like this thread, consider sharing it or following my awesome coauthors @saanarkethayan (with a banger of a first-ever first-author paper) @gupta__abhay and @mansiege. Happy to answer questions in the comments. [11/11]
!.! @xypyth
44 Followers 4K Following
Pircok @Pircok667
1 Followers 189 Following Focused on investing in U.S. stocks, happy to discuss stock market trends.
Ari Dyckovsky @adyckovsky
2K Followers 684 Following PhD candidate @Princeton studying and building systems for collective progress.
DeniseRicardo @OBA9KUoGlbf0KA
9 Followers 431 Following
Mathew Jacob @mat_jacob1002
134 Followers 65 Following Incoming PhD @uwcse. prev @DbrxMosaicAI, @siebelschool
Shizhe Diao @shizhediao
4K Followers 2K Following Research Scientist @NVIDIA focusing on efficient post-training of LLMs. Finetuning your own LLMs with LMFlow: https://t.co/UTykmQBwFr Views are my own.
Mo @sha3bani
64 Followers 1K Following
Vleapui @Vleapui897633
50 Followers 1K Following
Walton Cormier @cormier58237
90 Followers 3K Following
Feng Yao @fengyao1909
1K Followers 635 Following Ph.D. student @UCSD_CSE | Intern @Amazon Rufus Foundation Model Ex. @MSFTResearch @TsinghuaNLP
Margot Roberts @margot82539
34 Followers 2K Following
BM building AI @BMAIengineer
81 Followers 4K Following University student. Trying to build. Networking. Interests in AI Research,startups,software,CS and emerging technologies.
Blarekux @Blarekux44895
42 Followers 1K Following
Aaron @Norapom04
712 Followers 259 Following leisuremaxxing my productivity using twitter | occasionally softmaxxing matrixes | always dilly dallying | views are not mine
Fernando Larson @LarsonFern63456
30 Followers 2K Following
Rishabh Singh @rishabhs
970 Followers 91 Following Research Lead @Databricks. Previously @Meta GenAI, Google Brain @GoogleAI, @MSFTResearch, @MIT_CSAIL @IITKgp
Rohan Paul @rohanpaul_ai
83K Followers 8K Following Compiling in real-time, the race towards AGI. 🗞️ Don't miss my daily top 1% AI analysis newsletter directly to your inbox 👉 https://t.co/6LBxO8215l
ravirajm @ravirajm
225 Followers 3K Following Tech-Travel-Food. From two cities by the bay-SF and Mumbai. Building next gen AI data center infra. Stumbled into SiPho. Ex AWS /Arm. Opinions are solely mine
Nythew @NythewRrY
30 Followers 1K Following
William Whistler @willwhistler
317 Followers 1K Following Researcher in computational complexity and proof assistants. Still occasionally a reverse engineer.
Clayton ✪ @claytonwtx
206 Followers 3K Following
Bret Grinslade @bretgr
231 Followers 2K Following Data Analytics and Insights Product Manager @Oracle. #OracleAnalytics #Oracle #ML. Love to talk games as well! Views are my own.
Soumith Chintala @soumithchintala
250K Followers 1K Following Cofounded and lead @PyTorch at Meta. Also dabble in robotics at NYU. AI is delicious when it is accessible and open-source.
Clark Benham @ClarkBenham2
30 Followers 894 Following
Tyler Griggs @tyler_griggs_
554 Followers 349 Following CS PhD student @UCBerkeley Sky Lab, co-leading @NovaSkyAI and building SkyRL | Previously @GoogleCloud infra | @Harvard 2020
Irwan Bello @IrwanBello
7K Followers 3K Following Supercomputers & Friends AGI research & products founding team @reflection_ai ex @OpenAI, founding team @character_ai
Raymond Ng @Raymondng_aisg
8 Followers 1K Following
Chris Klaus @cklaus1
3K Followers 5K Following CEO of Fusen. Connecting students with mentors, investors, and funding opportunities through our Fusen accelerators. @cklaus.bsky.social on Bluesky.
Dung Doan @dungdx34
332 Followers 7K Following
Oskari Ajanki @OskariAjanki
30 Followers 5K Following
Eva Louise Marie Gabr... @e681554349
11 Followers 7K Following
Brad Neuberg @bradneuberg
12K Followers 9K Following Staff Machine Learning engineer @planet. Prev @ Dropbox & Google. Started coworking. Interests: ML, space, Earth Observation, VR. https://t.co/m7fXSRYQW3
Arjun Narayan @narayanarjun
4K Followers 1K Following Boris Babayan said "architecture, operating system and languages, compiler, it's only one project". Investing in improvements to the project @AmplifyPartners.
cresslank @cresslank
20 Followers 604 Following
Nidal @imleslahdin
2K Followers 1K Following What's the Kolmogorov Complexity / Minimum Description Length of a Reasoning Language Model? LLMs, AI/ML, Data Science. PhD student. 🇹🇳➡️🇺🇸
Forrest Cox @tetsuotrees
945 Followers 6K Following at play in the intersection of lifesci and of finance (and trying to not get run over) | opinions = mine
Sean Kirmani @SeanKirmani
3K Followers 547 Following Research @OpenAI. Interested in intelligence, understanding, and science.
Sarah Catanzaro @sarahcat21
14K Followers 1K Following “All methods are sacred if they are internally necessary” (GP @amplifypartners, prev @canvasvc; Head of Data @Mattermark; @palantirtech; @c4ads)
Ari Dyckovsky @adyckovsky
2K Followers 684 Following PhD candidate @Princeton studying and building systems for collective progress.
Rohan Varma @rvarm1
1K Followers 478 Following Research Engineer @Meta Superintelligence working on scaling. Former @PyTorch core developer
Zachary Charles @MatharyCharles
1K Followers 408 Following distributed machine learning @ google | sometimes mathematician
Wanchao Liang @wanchao_
1K Followers 225 Following building @thinkymachines ex-PyTorch @ Meta. Author of PyTorch DTensor and TorchTitan. Opinions are my own
NovaSky @NovaSkyAI
3K Followers 15 Following Next-generation Open Vision and AI @BerkeleySky Contact: [email protected]
Kilian Lieret @KLieret
882 Followers 40 Following Research Software Engineer at Princeton University. AI agents & benchmarks for software engineering.
Stuart Sul @stuart_sul
1K Followers 123 Following ml research @cursor_ai, cs @Stanford, mlsys @HazyResearch
Pratyush Maini @pratyushmaini
3K Followers 468 Following Data Quality x Privacy | PhD @mldcmu | Founding Team @datologyai | BTech @iitdelhi
Mathew Jacob @mat_jacob1002
134 Followers 65 Following Incoming PhD @uwcse. prev @DbrxMosaicAI, @siebelschool
Shizhe Diao @shizhediao
4K Followers 2K Following Research Scientist @NVIDIA focusing on efficient post-training of LLMs. Finetuning your own LLMs with LMFlow: https://t.co/UTykmQBwFr Views are my own.
Casey Handmer @CJHandmer
54K Followers 4K Following Physicist, Immigrant, Pilot, Dad. Former Caltech, Hyperloop, NASA JPL. Founder @terraformindies. Read scrolls. Build more solar!
Zhiqing Sun @EdwardSun0909
19K Followers 1K Following Agents @Meta MSL TBD Lab. previously posttraining research @OpenAI train LLMs to do things: deep research, chatgpt agent, etc. CS PhD @LTIatCMU
Edward Z. Yang @ezyang
14K Followers 1K Following I work on PyTorch at Meta. Chatty alt at @difficultyang.
Brandon Cui @BrandoCui
63 Followers 136 Following
Ethan He @EthanHe_42
15K Followers 815 Following AI @xai | prev @nvidia @AIatMeta @CarnegieMellon | 8k citations 5k GitHub stars | views are my own
Sayash Kapoor @sayashk
10K Followers 2K Following CS PhD candidate @PrincetonCITP. I tweet about AI agents, AI evals, AI for science. AI as Normal Technology: https://t.co/5amOkqKDf2 Book: https://t.co/DabpkhNrcM
Junhua Mao @junhuamao
904 Followers 81 Following Lead personality and model behavior research @OpenAI; Previously built the object understanding system and foundation models for self-driving @Waymo
Elaine Ya Le @ElaineYaLe6
5K Followers 146 Following AI researcher @OpenAI | ex @Google Brain / @GoogleDeepmind | @Stanford PhD
Ivan Zhou @ivanzhouyq
1K Followers 436 Following AI research engineer @Databricks 🧱 Prev @Uber AI, @StanfordCRFM, @LandingAI. I love computer vision in many ways 📸👨🏻💻🌁
Feng Yao @fengyao1909
1K Followers 635 Following Ph.D. student @UCSD_CSE | Intern @Amazon Rufus Foundation Model Ex. @MSFTResearch @TsinghuaNLP
DANΞ @cryps1s
14K Followers 445 Following CISO @OpenAI | Ex-CISO @PalantirTech | Occasional Shitposter | 🇺🇸 All views are my own, not my employer. Duh. (Tweets == 30d retention)
Aviral Kumar @aviral_kumar2
5K Followers 355 Following Assistant Professor of CS & ML at @CarnegieMellon. Part-time Research Scientist Google. PhD from UC Berkeley.
Sanmi Koyejo @sanmikoyejo
3K Followers 104 Following I lead @stai_research at Stanford. Co-founder @VirtueAI_co
Owain Evans @OwainEvans_UK
16K Followers 357 Following Runs an AI Safety research group in Berkeley (Truthful AI) + Affiliate at UC Berkeley. Past: Oxford Uni, TruthfulQA, Reversal Curse. Prefer email to DM.
Dylan Foster 🐢 @canondetortugas
3K Followers 1K Following Foundations of RL/AI @MSFTResearch. Previously @MIT @Cornell_CS https://t.co/vQIdUzsw8B RL Theory Lecture Notes: https://t.co/bhgL3aKIk0
David Pfau @pfau
29K Followers 2K Following Knowledge manifests itself in radiant dreams that shimmer like the wild sun Views are my own https://t.co/xqtVHHVI17 on 🦋
davidad 🎇 @davidad
20K Followers 9K Following Programme Director @ARIA_research | accelerate mathematical modelling with AI and categorical systems theory » build safe transformative AI » cancel heat death
Rohan Pandey @khoomeik
38K Followers 2K Following descending cross-entropy to ascend entropy || prev research @OpenAI @CarnegieMellon '23
Bob McGrew @bobmcgrewai
28K Followers 1K Following Learning new things. Former Chief Research Officer at OpenAI, early exec at Palantir, early employee at Paypal.
kalomaze @kalomaze
18K Followers 2K Following ML researcher (@primeintellect), speculator • extremely silly jester
Casey Flint @FlintCasey
3K Followers 791 Following 🇦🇺🏄♀️ testimonials include “oddly calibrated” and “the right kind of weird”. @reflection_ai
gabriel @GabrielPeterss4
35K Followers 488 Following research sora at @OpenAI, previously at midjourney, swedish high school dropout
Prem Qu Nair @premqnair
5K Followers 910 Following @cognition, previously @nuro @princeton. Pursuing 70mm, 225lb, and $0.10/piece
Yang Song @DrYangSong
14K Followers 940 Following Leading Strategic Explorations @OpenAI. Score-Based / Diffusion Models. Consistency Models. Optimization & Architecture.
Chen Sun 🤖🧠🇨... @ChenSun92
2K Followers 398 Following Research Scientist @ Google DeepMind Building memory & open-ended AI ex-neuroscientist ex-IMO team Canada Views are mine alone not GDM's.Trends for United States
You might like
