PhD student @UniofOxford and @MetaAI, prev Renmin University of China (RUC)
Oxford, UK
Joined September 2022
- Tweets255
- Following716
- Followers284
- Likes1.3K
Tingchen Fu retweeted
google scholar 👉 x scholar
wenhaochai.com/pages/x-schol…
Tingchen Fu retweeted
proceedings.neurips.cc/paper…
All credit to my student @jrosseruk for this prescient paper on the safety of multi-agent swarms from nearly 2 years ago.
Tingchen Fu retweeted
We discovered and communicated with a Chinese AI agent fleet. We believe this is the first time that researchers have ever made contact with a rogue agent fleet in the wild.
Nearly all prior public agent incidents (like the HuggingFace hack) came from American labs. These agents were hosted by Tencent's evaluation research team, and they targeted Alibaba's Amap.
Tingchen Fu retweeted
1/ New community checkpoint for integer multiplication:
κ = 0.00161272137 ≈ 2⁻⁹·²⁷⁶
T(n) = O(n(log n)⁰·⁹⁹⁸³⁸⁷²⁸)
Reviewed and merged to main. A cumulative community result
Verified and merged PR 210 from @dysmemic which pushes the frontier to
κ ≈ 2⁻¹⁰·⁴⁶ (tightened from κ = 2⁻¹⁸²)
This builds on contributions from Dugongue, @cfky_, Henry Grant, @RohanArun, Jacob Sussman, @avi_eisen , icekylinx, Romyxen, Evan McKinney and many others.
Since our last update, the community has increased our conditional integer-multiplication exponent saving by 54%, from κ ≈ 0.000460917 to κ ≈ 0.000710046. This combines several contributors’ circuit, memory-reuse and geometric improvements—and substantial new verification work.
Tingchen Fu retweeted
👋 𝗛𝗲𝗹𝗹𝗼 𝗧𝘄𝗶𝘁𝘁𝗲𝗿! 𝗙𝗶𝗿𝘀𝘁 𝘁𝘄𝗲𝗲𝘁 𝗮𝗹𝗲𝗿𝘁 🚨
Why not give it to a paper!
Excited to share our work at @Meta: 𝗣𝗿𝗼𝗪𝗔𝗠
In this paper, we try to answer:
"How should a robot imagine the future before it acts?"
🌐 Webpage: sii-ferenas.github.io/ProWAM…
🧵👇
Tingchen Fu retweeted
Looking for research engineer contractors for our team, working on AI Research Agents (AIRA). We are building agents that can perform diverse research tasks, run their own experiments, share knowledge and self-improve. Apply here: uk.meta.talentnet.community/…
Tingchen Fu retweeted
4B parameters. ZERO distillation. 61.5% on SWE-bench Verified 🤯
Meet FrogNano 🐸: Qwen3.5-4B post-trained purely with RL on synthetic tasks from TaskPilot.
Just 5 iterations × 300 tasks.
Who said coding agents have to be huge? 🐸
Tingchen Fu retweeted
Is RL optimizing the right objective? 🤔
Should we maximize mean reward? Best-of-k? Which k?
Standard RL pulls on the mean and often the distribution collapses to a spike. The tail dies 🥲
We introduce Tail-Likelihood Reinforcement Learning (TailRL).
It maximizes the mean reward while simultaneously maximizing coverage over high reward outputs.
🧵 1/n
Tingchen Fu retweeted
来个暴论+干货：A➗根本就不懂 Harness.
在harness上，和OpenAI对比起来，A➗更像一个营销公司。
Codex Desktop 完全领先 Claude Desktop 一个时代。
●codex desktop 的UI UX被所有同行学习，定义了agent GUI的设计规范
●扎实优雅的架构设计。真正的技术扎实和设计优雅。老早就大幅度重构，实现了统一 runtime host: V2 使用了 app-server 架构，全端统一使用一个runtime。对比之下，CC 架构到现在还是一个 session 一个 bun，天天堆代码屎山。而codex做到了，甚至你可以用自己的Client连接别人部署的 Codex App Server。
A➗还老吹自己是，Infra工程师特别多。我怎么感觉你根本就是营销公司？你的设计和架构在哪呢？
●各种Agent harness benchmark测试基本都是效果最好的一批，不管是Maka做的Terminal-Bench 测试还是Runta 的 Agent harness 测试，都是最强的那一批。
●在Astra上更是实现了异步工具调用与后训练，边执行边推理，实现了非阻塞调用，大幅度加快了执行，能明显感觉到现在执行时间短了。
●Codex老早就实载了服务端侧的上下文压缩，所以Codex虽然上下文要短，但是实际上效果表现却非常好。
在Astra上更是实现了异步工具调用与后训练，边执行边推理，实现了非阻塞调用，大幅度加快了执行，能明显感觉到现在执行时间短了。
●Responses API 做了 mid-turn steering，实现了动态交互调整，在任务进行中可以插入新指令或中间修改要求，直接把穿插修改放到了底层的 API层。
全球第一的 Computer Use 的的能力更是不用多说了，大家都吹爆了。
●Codex 还做了上下文管理的前沿探索，Context Windows 也已经实现在Codex里面了。用上下文窗口的滑动，而不是用压缩。这样子就通过可搜索历史主动换窗，让长期的信息能够沉淀，可以被召回。
●Codex的Memory目前也是越做越好了，基本上是目前实现里面最无感的一家。并且实现了dreaming机制。
●实装的 PTC / Code Mode 机制配合 astra 强大的后训练能力。让模型的能力大幅度跃升。是真正的Harness和模型相互配合的典范。
●Claude团队设计 MCP也是埋了不知道多少坑，根本没有借鉴历史很多协议的设计经验和精华。Plan mode也是Claude设计的，现在已经成了历史糟粕了。
Tingchen Fu retweeted
I've written 70+ recommendation letters for students applying graduate programs in my career. This is the saddest one I have to redo.
// A former undergrad student got denial of entrance with a 5-year ban, when flying to the US for PhD study. As usual, no reason was given. I can't imagine what he have gone through.
Tingchen Fu retweeted
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set.
AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools.
We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
Tingchen Fu retweeted
AI research agents can generate hundreds of ideas in seconds, but evaluating each can take days of GPU time. When compute is limited, which idea deserves execution?
We introduce AI Research Preference Models (RPMs) to assess ideas & focus compute on the most promising paths 🧵👇
Scores on DeepSWE ProgramBench
PosttrainBench Horizonmath
are more than doubled than the Hy v3
Glad to see these benchmarks aren't solved yet especially with how quickly we're burning through new ones.
Introducing Hy4 preview
◆ Open weights: 770B MoE, 49B active, 1M context
◆ Built for real work: code, docs and analysis, scientific research
◆ Hy4 preview helped optimize its own training pipeline, kernels included
◆ We ship upgrades, not price hikes
Free in @WorkBuddy_AI for the next two weeks
Blog：hy.tencent.ai/research/hy4-p…
GitHub：github.com/Tencent-Hunyuan/H…
Get WorkBuddy：workbuddy.ai/
Tingchen Fu retweeted
Can frontier AI agents achieve Recursive Self-Improvement by turning a weak method into one that performs better on hidden data?
Introducing RSI-Exam 🧵: 88 executable research tasks across 6 domains, now open for task contributions from every field🧑🔬.
Its task bank spans virtual cells, TPU kernels, chip design, quantitative finance, agent harnesses, model distillation, and much more.
Across these tasks, every rollout follows the same core protocol:
🔁 An agent inherits a working method and iterates using visible data.
🔒 Only its final artifact enters a fresh-container evaluation on the hidden set.
📊 The result so far: Opus 5 leads the 88-task leaderboard with a mean hidden-set score of 0.464.
📢 Want to help build the next release? Author an executable task, review one, or audit a rollout. Substantial contributions qualify for paper authorship.
Special thanks to @XinyeYee for the tremendous effort in leading this project.
Explore: rsi-exam.ai/
Blog: rsi-exam.ai/blog.html
Contribute: rsi-exam.ai/contribute.html
GitHub: github.com/aiming-lab/RSI-Ex…
Tingchen Fu retweeted
TL;DR: This is one of the most important and exciting opportunities in AI on the planet - please read on.
The British Open-ended Learning & Discovery Lab is creating the perfect place for paradigm breaking AI research in the name of open-source and open-science. We have agency, we funding, we have unprecedented amounts of compute*, but WE NEED YOU!
..and we have created the dream job for you: The BOLD Fellow. This job combines a fast-moving, high agency, collaborative environment with full academic freedom and a salary that pays the bills.
Apply by noon UK time on the 15th of September for this once in a lifetime opportunity to shape the history of our field and of our planet:
my.corehr.com/pls/uoxrecruit…
*by academic standards
Tingchen Fu retweeted
Good morning from Vienna People of Pi 🌞
Sunday meditations from @badlogicgames and @mitsuhiko
- On memory: code is the truth
- Bash is all you need
- Build context efficient tools
Tingchen Fu retweeted
🪄 MAGI-2 Preview: The Open-Source 100B+ Ultra-Fine-Grained MoE Video Generation Model
📄 Tech Report: sand.ai/blog/magi-2-preview
💻 GitHub: github.com/SandAI-org/MAGI-2…
🏆 114B total parameters, with only 6B activated per token for top-tier performance
🔓 100% open-source & tech report
💰 1/10 the generation cost of mainstream models
⏳ 1/7
Tingchen Fu retweeted
Excited to share what we have been cooking in the past few months: Muse Glimmer: research.meta.ai/blog/introd…
