-
Tweets56
-
Followers282
-
Following392
-
Likes458
Introducing VaSE: Value-Aware Stochastic KV Cache Eviction. Reasoning models think in CoT, bloating the KV cache. Eviction caps memory but suffers capability drop. VaSE is a training-free recipe that cuts that cost: keep large-magnitude value states, evict stochastically.
Excited to share that I’ve started @GoogleResearch as a student researcher today. I'll be working on tabular foundation models. Come and chat if you are around at Google or at the Bay Area.
Excited to share that I've started my summer internship at SystemsResearch@Google in Sunnyvale, working on agentic environment generation! Always happy to chat about coding agents or LLM memory too. If you're around the Bay Area, would love to meet up.
The future risk of computer-use agents won’t come only from malicious prompts. It will come from agents that can flawlessly follow normal instructions straight into harm. Introducing 𝐎𝐒-𝐁𝐥𝐢𝐧𝐝: a realistic but overlooked setting where every task begins with a benign user instruction, yet the harmfulness only emerges as the agent acts in the environment.
New paper: Convergent Evolution: How Different Language Models Learn Similar Number Representations. Language models, classical word embeddings, and even raw token frequencies all develop the same Fourier features for numbers. But only some develop the underlying structure. 🧵
After three papers on Fourier features in LLMs, I think there's a principle worth naming. How should we do science on an LLM? It corresponds to the existential questions: > who am I? ↔ the phenomenon. > where do I come from? ↔ the emergence. > where am I going? ↔ the use. 🧵
8/8 Both artifacts may find use beyond this paper. 🦣 BEHEMOTH as a testbed for diverse memory extraction approaches (self-evolving, routing-based, skill-based, and beyond). 🌱 CluE for any setting where one agent must handle heterogeneous demands, e.g. serving users with distinct habits. w/ @TengxiaoLiu, @BillJohn1235813, @taiwei_shi, @linxins2, @robinomial Check out the paper & code if this resonates!
7/8 Bonus findings: • CluE preserves strengths when starting from a stronger seed • Transfers to Gemini-3-Flash backend • Single-step gains carry over to continual memory settings • Produces clean, structured taxonomies, not bloated rule lists
🧵 1/8 What should an LLM assistant remember across conversations? Existing memory work studies this one task at a time. But real-world assistants see all kinds of conversations, and that changes the problem. Introducing BEHEMOTH 🦣 + CluE 🌱: a benchmark & self-evolving method for heterogeneous memory extraction. 📄 Paper: arxiv.org/abs/2604.11610
Frontier LLMs don't debug, they regenerate. We built PDB to measure that gap, GPT-5.1-Codex pass unit tests >76% of the time, but touch only <45% of the right lines. Even Claude Code touches only ~50%. 📄 Paper: arxiv.org/abs/2604.17338 🌐 Project: precise-debugging-benchmark.github.io
Coding agents running 24/7 will unlock a lot of breakthroughs 🚀. Easy to feel like we're being replaced 😨. But the real question: What can we learn from this, and where do they still fall short? New blog ⬇️
Auto research is on 🔥 We give algorithmic problems (like circle packing) to general coding agents, let it run overnight. 🌙 Agents reach SoTA. But more importantly: we analyze 100+ hours of trajectories to understand how it gets there 🧵
A practical insight!
🏧Giving your agent unlimited tool calls doesn't make it smarter. 💡Why? It lacks 'Budget Awareness'! Introducing Budget Tracker, a simple plug-in that enables more effective scaling behaviors: higher performance, lower cost. Paper: arxiv.org/pdf/2511.17006
Announcing 🔭✨Hubble, a suite of open-source LLMs to advance the study of memorization! Pretrained models up to 8B params, with controlled insertion of texts (e.g., book passages, biographies, test sets, and more!) designed to emulate key memorization risks 🧵
# 🚨 4B open-recipe model beats Claude-4-Opus 🔓 100% open data, recipe, model weights and code. Introducing Polaris✨--a post-training recipe for scaling RL on advanced reasoning models. 🥳 Check out how we boost open-recipe reasoning models to incredible performance levels (65 → 79 on AIME25) through RL training on open-source data and academic-level resources. 📑Notion: honorable-payment-890.notion.site/POLARIS-A-POst… 📗Blog post: hkunlp.github.io/blog/2025/Pola… 🤗Model & data: huggingface.co/POLARIS-Project 💻Code: github.com/ChenxinAn-fdu/…
Juan Weber @juanweber76
8 Followers 401 Following
Jacob Choi @jacob_jchoi
30 Followers 286 Following incoming phd @sbucompsc, mscs @csatusc, previously @emorynlp
ChengSong Huang @ChengsongH31219
316 Followers 174 Following @WUSTL CSE PhD candidate, Student Researcher @GoogleAI @googlecloud . R-Zero, Lorahub, Self-Calibration Previous research intern @SeaAIL @TencentGlobal
Siddartha Devic @sid_devic
342 Followers 636 Following PhD student @CSatUSC, previously intern @apple @amazon, ug @UT_Dallas. Trustworthy AI and ML.
- MrProtein ( guy who... @AlexHeisenberg_
3 Followers 7K Following Hilbert Space 📍Don’t follow back : I don’t have a line to write here ! builds softs at lab :) Bouncing electron; Interest; nuclear physics prev: YC , AWS Cloud
Laura Gomezjurado @LauraPGomez_
209 Followers 1K Following Robust & Trustworthy AI @stanford | Prev. @MSFTresearch | https://t.co/UHLf9HNGq8
Prakhar Kothari @PrakharKo
13 Followers 1K Following Deep Learning student | MBAi @ Kellogg School of Management | Physics enthusiast | SDE | Electrical Eng @ IIT Patna | UPSC CSE & IFS Interview stage
Harshitha M @harshithamanju3
100 Followers 2K Following Multimodal LLMs & UAV/Edge Intelligence | PhD Researcher at UT Dallas | HPC-AI on GPUs & Supercomputing | Passion for AI | ex-BMW | ex-Deloitte
Ashish @ashishjayamohan
2K Followers 716 Following Member of Technical Staff @perplexity_ai | UCSD | Systems Saturday
David Cai @davidjiyangcai
643 Followers 6K Following MRes/PhD @LSEEcon & @LSEManagement Prev: Pre-Doc @ ETH Zurich (@eth_cle). Undergrad @RiceUniversity
Guohao Li 🐫 @guohao_li
14K Followers 4K Following Founder @Eigent_AI / @CamelAIOrg. Scaling RL Environments for Agents. Prev Oxford, KAUST, ETHz, Intel, Kumo.
Ali Ihsan Nergiz @ICM... @ainergiz
1K Followers 1K Following founding engineer at delfa, building the ai layer of clinical trials. medical doctor
Gaotang Li @GaotangLi
225 Followers 354 Following Ph.D. @UofIllinois | Undergrad @UMich. Language Models.
invi-bhagyesh 🏴... @invi_bhagyesh
652 Followers 7K Following maths ∩ AISafety, sophomore, I play dice, working w/ @lionellevine, currently on exchange @USTC
de jia @dejia49220082
32 Followers 3K Following
Arshia Arya @AryaArshia
2K Followers 6K Following PhD student @ucsd_cse. Ex @MSFTResearch. Alum @bitspilaniindia
Ralphael Lundt @RLundt20993
56 Followers 7K Following
conan kinde @ConanKinde
0 Followers 6K Following
Miaosen Chai @miaosenc
5 Followers 50 Following Incoming PhD @UChicago | Investor @sequoia | prev @nlp_usc Research & Build LLM
Sitao Cheng @TonyCheng990417
276 Followers 662 Following Phd Student @UWCheritonCS @VectorInst. Reasoning and Generalization of Language Agents. Ex @ucsbNLP, @MSFTResearch, @NanjingUnivers1.
Qingchuan (Tony) Yang @qcyang20xx
96 Followers 277 Following PhD at @CSatUSC. B.S. 24' & M.S. 25' at @UChicago. @ProphetArena
Mengxi Wu @MengxiWu567
19 Followers 91 Following Ph.D. Student in Machine Learning @CSatUSC @USCMathDept. Previously: @nyuniversity @UMichECE
Reza (Rey) Sanayei @r_sanayei
36 Followers 895 Following ML Research @ScaleAILabs | MSCS @Uarizona | NLP @LabCLU
Wiktoria Leks @wiki_answerz
377 Followers 3K Following Researcher in green solutions, longevity and access to GenAI tools.
Kyoung Whan Choe @kywch500
1K Followers 2K Following Robot Learning Engineer @ https://t.co/wcLx79rCuW
Epsilon Guanlin Lee @Epsilon_Lee
320 Followers 4K Following PhD, MLer, CLer (NLPer), ML Engineer at https://t.co/gX6Lem59Co, have belief in interpretability research of AI/ML/NNs
Thomas Kleine Buening @thomasklbg
260 Followers 261 Following Postdoc @ETH @ETH_AI_Center, prev. @UniofOxford @turinginst @unioslo, reinforcement learning
Muni Kumar @sssvmkumar
120 Followers 5K Following
_ @_prmd_
142 Followers 5K Following Product Developer @ SAP, Product Security, AppSec, Data Privacy, GDPR. Tweets are personal. RTs aren't endorsements.
locke @Yangyue_Wang
64 Followers 750 Following founding research engineer @figbrains & oss researcher @ManifoldRG Exploring open-ended multi-agent complex systems
Yuecheng Li @Yuecheng_Lee
286 Followers 1K Following Researcher @ Kuaishou Technology; CS MSc @ SYSU Prev @alibaba_cloud @NetEaseGames_EN Work on #LLM (LLM4Rec, Reasoning) and #Trustworthy_AI (LLM Eval, Privacy)
Momo @th000fly100
39 Followers 5K Following
Harness Engineering @harnessengr
250K Followers 11K Following The more things change, the more they stay the same. DM for repost
superops @xidianwlc
60 Followers 3K Following 一个vibe coding的探索者，在调动各个AI厂商的AI接口构建最地道的markdown 编辑器，可以github搜索markrender就可以找到它
Jieyu Zhao @jieyuzhao11
4K Followers 963 Following Assistant Prof. @CSatUSC, @USC || Postdoc @ClipUMD || PhD from @UCLANLP, @UCLA. #NLP, #ML, #TrustworthyNLP
Hossein A. (Saeed) Ra... @srahmanidashti
1K Followers 3K Following PhD Student in Foundational AI @ucl_nlp @FAICDT1 @ai_ucl @uclcs @UCL. Enrichment Fellow at @turinginst
Joseph Hawa @Joseph___Hawa01
337 Followers 7K Following HSBC Deputy Manager at HSBC Bank Former Account Officer at HSBC Senior Financial Accountant From Jeddah, Saudi Arabia Lives in London, United Kingdom
Louis @Louis9687221579
75 Followers 3K Following Mainline Economics | Idea page | ramblings of a schizo
Wenyue Hua @HuaWenyue31539
2K Followers 865 Following senior researcher @MSFTResearch, AI Frontiers Postdoc @ucsbNLP Ph.D. @RutgersCS KAUST AI Rising Star LLM-based agent, LLM reasoning
Ankur Bohra @AnkurBohra9
52 Followers 5K Following
Miaosen Chai @miaosenc
5 Followers 50 Following Incoming PhD @UChicago | Investor @sequoia | prev @nlp_usc Research & Build LLM
Jiefeng Chen @jiefengchen1
857 Followers 742 Following Research Scientist at Google | Working on LLM Research.
Laura Gomezjurado @LauraPGomez_
209 Followers 1K Following Robust & Trustworthy AI @stanford | Prev. @MSFTresearch | https://t.co/UHLf9HNGq8
Siddartha Devic @sid_devic
342 Followers 636 Following PhD student @CSatUSC, previously intern @apple @amazon, ug @UT_Dallas. Trustworthy AI and ML.
Gaotang Li @GaotangLi
225 Followers 354 Following Ph.D. @UofIllinois | Undergrad @UMich. Language Models.
Ashish @ashishjayamohan
2K Followers 716 Following Member of Technical Staff @perplexity_ai | UCSD | Systems Saturday
Bowen Wang @BowenWangNLP
887 Followers 450 Following 2nd year Ph.D. student @HKUniversity, Prev. @Tsinghua_Uni. Cooking digital agents at @Alibaba_Qwen, Prev @Kimi_moonshot
Hongyu Li @Hongyu_Lii
1K Followers 973 Following Robotics researcher at @BrownCSDept 🐻 and @nvidia. Visiting @ColumbiaCompSci. Prev: @rai_inst, @AmazonScience, @Honda
Ray Yang @RuixinYang6
129 Followers 796 Following Incoming PhD student @UW | MSCS @ICatGT | Prev. undergrad @UBC_CS
Dennis Y Menn @MennDennis
145 Followers 878 Following Ph.D. @utexasece. Building real-time video generation on consumer GPUs!
Harness Engineering @harnessengr
250K Followers 11K Following The more things change, the more they stay the same. DM for repost
Thomas Kleine Buening @thomasklbg
260 Followers 261 Following Postdoc @ETH @ETH_AI_Center, prev. @UniofOxford @turinginst @unioslo, reinforcement learning
Jingming Zhuo @JingmingZhuo
109 Followers 1K Following M.S. at UW; B.S. from Jilin University. I work on agents and LLMs.
Mengxi Wu @MengxiWu567
19 Followers 91 Following Ph.D. Student in Machine Learning @CSatUSC @USCMathDept. Previously: @nyuniversity @UMichECE
Qingchuan (Tony) Yang @qcyang20xx
96 Followers 277 Following PhD at @CSatUSC. B.S. 24' & M.S. 25' at @UChicago. @ProphetArena
Yuecheng Li @Yuecheng_Lee
286 Followers 1K Following Researcher @ Kuaishou Technology; CS MSc @ SYSU Prev @alibaba_cloud @NetEaseGames_EN Work on #LLM (LLM4Rec, Reasoning) and #Trustworthy_AI (LLM Eval, Privacy)
Sitao Cheng @TonyCheng990417
276 Followers 662 Following Phd Student @UWCheritonCS @VectorInst. Reasoning and Generalization of Language Agents. Ex @ucsbNLP, @MSFTResearch, @NanjingUnivers1.
Arshia Arya @AryaArshia
2K Followers 6K Following PhD student @ucsd_cse. Ex @MSFTResearch. Alum @bitspilaniindia
de jia @dejia49220082
32 Followers 3K Following
Jiayi Geng @JiayiiGeng
2K Followers 325 Following PhD @LTIatCMU | Prev MSE @Princeton @PrincetonPLI, BS @mcgillu. Working on Multi-agent / LLM evaluation.
Xiangyi Li @xdotli
5K Followers 2K Following your friendly neighborhood eval guy, creator of SkillsBench ClawsBench chat about data, evals, rl environments, skills https://t.co/Jl1qzLItZn
Sakana AI @SakanaAILabs
73K Followers 0 Following Sakana AI is an AI R&D company based in Tokyo. Try Sakana Chat → https://t.co/1m2lSgnfB2
Wenyue Hua @HuaWenyue31539
2K Followers 865 Following senior researcher @MSFTResearch, AI Frontiers Postdoc @ucsbNLP Ph.D. @RutgersCS KAUST AI Rising Star LLM-based agent, LLM reasoning
Joongwon Kim @danieljwkim
714 Followers 324 Following PhD student @uwcse @uwnlp | Currently at @AIatMeta | Former undergrad @Penn
Taiwei Shi @taiwei_shi
2K Followers 467 Following AI Researcher & Ph.D. student @nlp_usc. Intern @MSFTResearch. Formerly @GeorgiaTech @USC_ISI. NLP, RL & Computational Social Science.
Polina Kirichenko @polkirichenko
4K Followers 1K Following Research Scientist at FAIR @AIatMeta & visiting researcher at Princeton @VisualAILab prev. PhD at New York University 🇺🇦
Yahan Li @YLiiiYLiii
112 Followers 152 Following PhD Student @USC Previously Master Student @JHUCompSci + Research Assistant @jhuclsp + CSE Undergrad @UMich. Interested in Clinical NLP and AI for Healthcare☀️
Jaeyoung Lee @lee__jaeyoung
150 Followers 679 Following Doing research on MLLM/LLM | Automated/Mechanistic Interpretability
Fangcong Yin @fangcong_y10593
311 Followers 775 Following CS PhD Student @NYU studying NLP. Prev: @Cornell_Bowers @UTAustin
Benhao Huang @huskydogewoof
1K Followers 764 Following Attracted by Loop Models➰and World Models 🌎 | M.S. student @mldcmu, Prev. @sjtu1896 | Opinions approved by my puppy.
Xi Ye @xiye_nlp
3K Followers 428 Following I study NLP. Postdoc fellow @PrincetonPLI. CS PhD @UTAustin.
Hao Zhu @_Hao_Zhu
3K Followers 1K Following Building the AI social brain for humans @StanfordNLP PhD @LTIatCMU
Jiahe Jin @jiahe_Jin0123
119 Followers 183 Following Undergrad @ SJTU, intern @LTIatCMU, supervised by @XiongChenyan. Previously intern @ GAIR, supervised by @stefan_fee
Dongwei Jiang @Dongwei__Jiang
542 Followers 825 Following Working on RL and agents at AWS. Previously reasoning and self-improvement. Six years of prior industrial research experience in speech processing and NLP
Fangyuan Xu @brunchavecmoi
594 Followers 726 Following 许方园👩🏻💻phd student @ nyu, interested in natural language processing
Harvey Yiyun Fu @harveyiyun
144 Followers 145 Following Ph.D. student @uchicago cs | Previously @nlp_usc
Bhavya Vasudeva @bhavya_vasudeva
169 Followers 566 Following PhD candidate @CSatUSC | BTech @iitroorkee'20 | ML/DL theory, Science of DL/LMs, Optimization | she/her
Ameya P. @AmyPrb
591 Followers 672 Following Postdoc @bethgelab; Previously: @OxfordTVG, @intelailabs Profile - https://t.co/To9NNR5Izc
Jiewen Hu @Jiewenhu02
9 Followers 40 Following
Tian Jin @jintian
583 Followers 486 Following PhD student @MIT_CSAIL, previously @IBMResearch, @haverfordedu .
Naomi Saphra @nsaphra
11K Followers 1K Following Waiting on a robot body. All opinions are universal and held by both employers and family. Now a dedicated grok hate account. Accepting ML/NLP PhD students.You might like
