Stephen Xie @stephenx_
@berkeleynlp @berkeley_ai | eecs & business @ucberkeleymet Berkeley & SF Joined August 2020-
Tweets44
-
Followers135
-
Following622
-
Likes253
new post. there's a lot in it. i suggest you check it out
Yup :) here are the reasoning traces from running gpt-oss 20b on the s1k dataset huggingface.co/datasets/Steph…
Yup :) here are the reasoning traces from running gpt-oss 20b on the s1k dataset huggingface.co/datasets/Steph…
New paper & surprising result. LLMs transmit traits to other models via hidden signals in data. Datasets consisting only of 3-digit numbers can transmit a love for owls, or evil tendencies. 🧵
🚀 Hello, Kimi K2! Open-Source Agentic Model! 🔹 1T total / 32B active MoE model 🔹 SOTA on SWE Bench Verified, Tau2 & AceBench among open models 🔹Strong in coding and agentic tasks 🐤 Multimodal & thought-mode not supported for now With Kimi K2, advanced agentic intelligence…
With Grok-4, RL is the new pre-training
With Grok-4, RL is the new pre-training
People are racing to push math reasoning performance in #LLMs—but have we really asked why? The common assumption is that improving math reasoning should transfer to broader capabilities in other domains. But is that actually true? In our study (arxiv.org/pdf/2507.00432), we…
Recently, there has been a lot of talk of LLM agents automating ML research itself. If Llama 5 can create Llama 6, then surely the singularity is just around the corner. How can we get a pulse check on whether current LLMs are capable of driving this kind of total…
New paper: What happens when an LLM reasons? We created methods to interpret reasoning steps & their connections: resampling CoT, attention analysis, & suppressing attention We discover thought anchors: key steps shaping everything else. Check our tool & unpack CoT yourself 🧵
What Makes a Base Language Model Suitable for RL? Rumors in the community say RL (i.e., RLVR) on LLMs is full of “mysteries”: (1) Is the magic only happening on Qwen + Math? (2) Does the "aha moment" only spark during math reasoning? (3) Is evaluation hiding some tricky traps?…
Can we actually control reasoning behaviors in thinking LLMs? Our @iclr_conf workshop paper is out! 🎉 We show how to steer DeepSeek-R1-Distill’s reasoning: make it backtrack, add knowledge, test examples. Just by adding steering vectors to its activations! Details in 🧵👇
🔥 We introduce Multiverse, a new generative modeling framework for adaptive and lossless parallel generation. 🚀 Multiverse is the first open-source non-AR model to achieve AIME24 and AIME25 scores of 54% and 46% 🌐 Website: multiverse4fm.github.io 🧵 1/n
DeepSeek 671b and Qwen3 236b support with Megatron backend is now available as preview in verl v0.4.0 🔥🔥🔥 We will continue optimizing MoE model performance down the road. DeepSeek 671b: verl.readthedocs.io/en/latest/perf… verl v0.4: github.com/volcengine/ver…
Announcing OpenThinker3-7B, the new SOTA open-data 7B reasoning model: improving over DeepSeek-R1-Distill-Qwen-7B by 33% on average over code, science, and math evals. We also release our dataset, OpenThoughts3-1.2M, which is the best open reasoning dataset across all data…
RL scaling is here arxiv.org/pdf/2505.24864
Incredible work by my mentors and open-source collaborators—honored to have played a tiny part! Huge respect for Simon Huang & team for leading this! 👏🙏
Incredible work by my mentors and open-source collaborators—honored to have played a tiny part! Huge respect for Simon Huang & team for leading this! 👏🙏
🤯 We cracked RLVR with... Random Rewards?! Training Qwen2.5-Math-7B with our Spurious Rewards improved MATH-500 by: - Random rewards: +21% - Incorrect rewards: +25% - (FYI) Ground-truth rewards: + 28.8% How could this even work⁉️ Here's why: 🧵 Blogpost: tinyurl.com/spurious-rewar…
Giving your models more time to think before prediction, like via smart decoding, chain-of-thoughts reasoning, latent thoughts, etc, turns out to be quite effective for unblocking the next level of intelligence. New post is here :) “Why we think”: lilianweng.github.io/posts/2025-05-…
Jason @larget_jason
112 Followers 104 Following pioneering agent experience (ax). ex fintech product designer. now building @intracelabs
Rohan Sontakke @rohoosta
19 Followers 115 Following
Alice Wilson @PrOve_zdd
1 Followers 297 Following
Jeremi Nuer @JeremiNuer
120 Followers 207 Following CompE @ UCSB. Prev @Livermore_Lab. Robot Learning, Piano, Triathlons
Gajesh @gajesh
43K Followers 3K Following engineer, chaos agent @eigen_labs ($20B+ AUM). fmr: @solana, LZ, FTX, TG. 2x founder. perpetually curious.
hayden @hxxwhite
104 Followers 189 Following Building test automation that thinks @tryrevyl // prev Purdue CS
Pivauk @Pivauk657
58 Followers 2K Following
Cerfu @Cerfu796
19 Followers 592 Following
Ankit Gupta @AnkitGuptaAI
1K Followers 1K Following researcher, alum @scsatcmu, ex @imctrading @facebook
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.
Huang Qichag @HQichag48397
427 Followers 8K Following
Jonas vasconceloss @Jonasvasco66460
152 Followers 4K Following
Veetob @Veetob1187
50 Followers 3K Following
Igor Kan @1gor_kan
5K Followers 7K Following ⊣|─ math, physics, stats, ai @UofT ─⊗─ building @lesenheit @styles_lab ─▭─ https://t.co/EuZcbXZ0wZ _/ _ philosophy, ancient near east lit.,history ⏚ :wq && exit
Teknium (e/λ) @Teknium1
50K Followers 5K Following Cofounder and Head of Post Training @NousResearch, prev @StabilityAI Github: https://t.co/LZwHTUFwPq HuggingFace: https://t.co/sN2FFU8PVE
May Wuckert @MayWuckert11451
28 Followers 2K Following
HUANG Qichang @huang85993
356 Followers 7K Following
Coohaw @Coohaw9357129
56 Followers 2K Following
Kevin Lu @_kevinlu
9K Followers 216 Following @thinkymachines. formerly: - @openai: RL, synthetic data, efficient models - @berkeley_ai: decision transformer, universal computation
Jerome Ku @KuJerome
58 Followers 5K Following
Erhawirhu @Erhawirhu018
12 Followers 885 Following
Keshav Badrinath @keshavbadrinath
2 Followers 6 Following cs @UOfIllinois | robotics | ai + mro @Robotoolworx
Yuetai Li @yuetai12575
225 Followers 570 Following Second year PhD @UW | Post-Training, LLM reasoning and synthetic dataset. https://t.co/cYAkbnCsCp Open to chat and collaborate!
Xian Zhao @0012ZhaoXian
23 Followers 381 Following
Tyler Griggs @tyler_griggs_
550 Followers 349 Following CS PhD student @UCBerkeley Sky Lab, co-leading @NovaSkyAI and building SkyRL | Previously @GoogleCloud infra | @Harvard 2020
Caleb Chan @calebyanchan
566 Followers 780 Following building Lance 🔨 your marketplace for everything | prev @google @salesforce & head of product @svc (acquired $XXXM) | @ucberkeley @joinodf
Brian Zhou @brian1zhou
494 Followers 2K Following Cofounded @a37_ai. On leave @Harvard, was @TJHSST.
Xohout @Xohout6352
26 Followers 2K Following
ayush @ayushrgarg
527 Followers 246 Following software engineering @uwaterloo | ex-swe @ https://t.co/cQZktqOkQ2, https://t.co/ULrtzSxmCu, @dainprotocol
Shane @shinheeshanelee
2K Followers 4K Following Co-founder @DeftRobo | systems integrator of wheeled humanoid robots for 🇺🇸 OEMs & Tiers 1-2 | prev @UCBerkeley, @Tesla Motor Design + Autopilot Hardware
Justin T Chiu @justintchiu
647 Followers 891 Following generating code at Cohere; phd in ml from Cornell; former Child
Anya Ji @__anyaj
87 Followers 146 Following CS PhD student @berkeleynlp / Previously @Cornell / NLP, cognitive science
Gatik @Gatik07116642
15 Followers 53 Following
Linxin Song @linxins2
368 Followers 522 Following PhD student @csatusc @nlp_usc | Master @waseda_univ | Research intern @SFResearch | NLP/CV, LLM/LMM, Agent, Trustworthy model, Synthetic data
Derek Xu @derekzxu
85 Followers 256 Following Eng @ Fireworks AI | @ucberkeleymet | @BerkeleyML | prev Ramp, TikTok, Salesforce, Built Robo.
Ananya Kumar @ananyaku
8K Followers 607 Following Researcher working on reasoning at @openai Previously PhD at Stanford University AI Lab with Percy Liang and Tengyu Ma
Sieboo @Sieboo9623064
6 Followers 20 Following
Gennady Korotkevich @que_tourist
20K Followers 1 Following
Jason Lee @jasondeanlee
18K Followers 4K Following Associate Professor at UC Berkeley. Former Research Scientist at Google DeepMind. ML/AI Researcher working on foundations of LLMs and deep learning.
Yuchen Jin @Yuchenj_UW
54K Followers 535 Following Co-founder & CTO @hyperbolic_labs 🧑🍳 fun AI systems. Previously: OctoAI (acquired by @nvidia) building @ApacheTVM, PhD @uwcse 🤖
andrew pignanelli @ndrewpignanelli
2K Followers 832 Following ceo @nycintelligence, i am going to get ai to run companies. timeline to agi is 12 months
General Intelligence ... @nycintelligence
605 Followers 0 Following The General Intelligence Company Of New York - Our mission is to enable the one person one billion dollar company
Qian Huang @qhwang3
14K Followers 331 Following prev @xai | CS PhD student @StanfordAILab (on leave)
Bryan Johnson @bryan_johnson
646K Followers 760 Following Conquering death will be humanity’s greatest achievement.
kache @yacineMTB
196K Followers 6K Following canadian startup founder. prev eng @ x, stripe. yacine_kv on insta i make my memes with https://t.co/pWRBfY8kn2 - I write a subscriber only blog. Subscribe!
Misha Laskin @MishaLaskin
15K Followers 214 Following Co-founder, CEO at @reflection_ai. Prev: Research @DeepMind. Gemini RL team.
Michael Luo @michaelzluo
541 Followers 213 Following Project Lead @Agentica_ | Prev. Researcher @GoogleDeepMind | PhD at UC Berkeley @berkeley_ai
Redwood Research @redwood_ai
1K Followers 6 Following Pioneering threat mitigation and assessment for AI agents.
John Schulman @johnschulman2
65K Followers 1K Following Recently started @thinkymachines. Interested in reinforcement learning, alignment, birds, jazz music
Rohan Sontakke @rohoosta
19 Followers 115 Following
Mohit Reddy @MohitReddy13
3K Followers 910 Following Understanding the universe @xAI; Previously: Co-Founder at @FennelAI, ex ML Infra at Google Brain, ex Infra at GCE Startups, Software, Tech, Infra, AI :)
Vigil Labs @vigillabs
1K Followers 2 Following
Stagehand 🤘 @Stagehanddev
3K Followers 32 Following Open-source, production-ready AI browser automations. 2M monthly downloads, built by @browserbasehq
Prophet Arena @ProphetArena
2K Followers 14 Following The AI benchmark for predictive intelligence, advancing collective foresight via human–AI collaboration, from SIGMA Lab @UChicagoCS @DSI_UChicago
Qianqian Wang @QianqianWang5
3K Followers 434 Following Postdoc at UC Berkeley and Visiting Researcher at Google. Former Ph.D. student at Cornell Tech. https://t.co/LyIdb5HmM9
DHH @dhh
554K Followers 131 Following Father of three, Creator of Ruby on Rails + Omarchy, Co-owner & CTO of 37signals, Shopify director, NYT best-selling author, and Le Mans 24h class-winner.
teddy. @tedddyoweh
12K Followers 475 Following curious mind / prev. @apple, @linkedin, @nasa mafia @shipfr8
marmik @marmikch
500 Followers 593 Following interpreter of latents, reader of tokens, listener of music | prev @GroqInc
Gajesh @gajesh
43K Followers 3K Following engineer, chaos agent @eigen_labs ($20B+ AUM). fmr: @solana, LZ, FTX, TG. 2x founder. perpetually curious.
Jeremi Nuer @JeremiNuer
120 Followers 207 Following CompE @ UCSB. Prev @Livermore_Lab. Robot Learning, Piano, Triathlons
Rich @Im_Rich_Zou
3K Followers 2K Following 20yo / growth @withdelphi Finance @northeastern / prev top 30 @leagueoflegends @parable_vc
Rachel Freedman @FreedmanRach
2K Followers 239 Following RLHF, LLMS, interpretability & safety | PhD researcher @berkeley_ai | Visiting researcher @Cambridge_Uni
Yuhuai (Tony) Wu @Yuhu_ai_
38K Followers 452 Following Co-Founder @xAI. Grok Reasoning, STaR, Minerva, AlphaGeometry, Autoformalization, AlphaStar, Memorizing transformer.
Machel Reid @machelreid
3K Followers 1K Following research scientist @googledeepmind ♊️ post-training/thinking/rl
hayden @hxxwhite
104 Followers 189 Following Building test automation that thinks @tryrevyl // prev Purdue CS
Jiayi Weng @Trinkle23897
3K Followers 142 Following MTS @openai, author of the entire post-training RL infra, core contributor of ChatGPT/GPT4/GPT4o etc. 30U30
swyx 🇸🇬 @swyx
125K Followers 3K Following achieve ambition with intentionality, intensity, & integrity - @smol_ai - @dxtipshq - @sveltesociety - @aidotengineer - @coding_career - @latentspacepod
Song Mei @Song__Mei
3K Followers 690 Following Assistant Professor at UC Berkeley, Department of Statistics and EECS. Researcher at OpenAI working on LLM training.
Myra Deng @myra_deng
1K Followers 136 Following aligning models @goodfireAI, prev @stanford and @twosigma
Joshua Batson @thebasepoint
4K Followers 677 Following trying to understand evolved systems (🖥 and 🧬) interpretability research @anthropicai formerly @czbiohub, @mit math
Runjin Chen @RunjinChen
503 Followers 52 Following Research Fellow @AnthropicAI | PH.D. student @UTAustin @VITAGroupUT | Previously BS/MS @sjtu1896
Lakshya A Agrawal @LakshyAAAgrawal
2K Followers 2K Following AI PhD @ UC Berkeley | GEPA Creator (https://t.co/EdPqvzj7k4) | Created https://t.co/YxPZsXZJeS | Past: AI4Code Research Fellow @MSFTResearch | Hobbyist Saxophonist
Rowan Zellers @rown
14K Followers 974 Following multimodal @thinkymachines. I also like to climb rocks and throw pottery. https://t.co/5Er4j39K71 (he/him)Trends for United States
You might like
