Bill Shen (@williamfshen) header
Bill Shen (@williamfshen) profile picture

Cambridge | ex-MSL | ex-Oracle

Joined March 2010
Photo by Bill Shen (@williamfshen)Photo by Bill Shen (@williamfshen)Photo by Bill Shen (@williamfshen)Photo by Bill Shen (@williamfshen)
Bill Shen retweeted
Anyone watching recursively self-improving agents knows how good they've gotten at optimising over a fixed benchmark with verifiable rewards, like math or coding. Our recent work, the Red Queen Gödel Machine (RQGM), co-evolves the evaluation itself together with the agent. 🧵
10
31
5
246
20,820
Excited to introduce our new paper, the Red Queen Gödel Machine (RQGM), that co-evolves evaluator with the self-improving agent. Prior self-improving agents keep smashing fixed benchmarks - but those saturate faster than we can build them, and struggles for open-ended tasks.
3
2
5
212
Fixed benchmarks got us this far. To go further, stop optimizing the yardstick - start co-evolving it.
29
We show RQGM delivers substantial gains: - Coding: beats the prior SOTA while using 1.35-1.72× fewer tokens - Paper writing: 1.78–1.86× higher acceptance - Olympiad-level proof grading: 9% higher ground-truth accuracy - Peer review: closing reviewer bias between AI and human work
32
Thanks @imperial_nlp for having me as an external speaker - really enjoyed the talk and the inspiring discussions. If you’re working on NLP / foundation models, I’d definitely recommend giving a talk here!
Replying to @imperial_nlp
We love hosting external speakers too, recent guests include @williamfshen (University of Cambridge) and @juand_r_nlp (the University of Texas at Austin). If you work in NLP / foundation models and would like to present your research, let us know and we'll make it happen!
2
3
661
📣 Our recent work at Meta: Rethinking Rubric Generation for Improving LLM Judge and Reward Modeling for Open-ended Tasks Rubric-based judging and reward modeling are powerful for open-ended tasks, but scaling requires controlled generation, not just free-form lists from an LLM.
12
24
2
134
14,999
Big thanks to my amazing host and mentor @iliasl and co-authors: @xinchiqiu, Chenxi Whitehouse, @LisaAlazraki, @ShashwatGoel7, @fvancesco, @TimonWilli, @akhilmathurs
5
396
📊 Key results: - Consistent gains across judges and datasets (e.g., up to +17.7 points on JudgeBench) - Using RRD for reward modeling yields stronger learning signals (boosting reward by up to +160%), with gains transfer to downstream performance, including high-stakes domains
5
387
To address this, we propose RRD (Recursive Rubric Decomposition), a principled framework for controlled rubric generation that recursively decomposes coarse rubrics into discriminative sub-criteria, filters misaligned and redundant dimensions, and reweights correlated signals.
7
398
Rubrics help LLM judge capture nuanced, multi-dimensional human preferences and to turn open-ended quality into RL rewards. However, rubric generation is often messy: poor coverage, conflated dimensions, misaligned preferences, and correlated criteria, hurting their performance.
6
510
Excited to share our new work from MSL 🚀 We show that small agents can scale to long-horizon, complex tasks via coordination. Strategy Auctions enable system-level gains: better task performance and cost–performance tradeoffs. Capabilities emerge from how agents work together!
📣 New paper from my Meta Superintelligence Labs internship: Scaling Small Agents Through Strategy Auctions. Small language models are cheap but don’t scale to long-horizon tasks. Strategy auctions help them punch above their weight. 🚀 (1/7) 🧵⬇️
2
163
📣 New paper from my Meta Superintelligence Labs internship: Scaling Small Agents Through Strategy Auctions. Small language models are cheap but don’t scale to long-horizon tasks. Strategy auctions help them punch above their weight. 🚀 (1/7) 🧵⬇️
8
31
5
190
17,406
Excited to share our work "LUNAR: LLM Unlearning via Neural Activation Redirection" We propose a novel method that achieves SOTA unlearning effectiveness and significantly improves retained model's controllability. Paper: arxiv.org/pdf/2502.07218 More in threads🧵 @xinchiqiu
4
3
6
400
(4/N) "Experiments" We show LUNAR delivers SOTA unlearning effectiveness and significantly improves retained model's controllability. It's also robust against various attacks and highly computationally efficient! Visualization of results:
1
58
(3/N) "Method" Grounded in the Linear Representation Hypothesis, LUNAR operates by redirecting unlearned representations to regions that trigger the model’s inherent ability to express its inability to answer.
2
46
(2/N): "Motivation" Existing methods lack 𝙘𝙤𝙣𝙩𝙧𝙤𝙡𝙡𝙖𝙗𝙞𝙡𝙞𝙩𝙮, often leading to hallucinated, gibberish or monotonous responses. Worse, they require extensive LR tuning, be vulnerable to attacks, and still impose non-negligible training costs - hindering wide adoption.
2
49
On Sunday at #KDD2024 (@kdd_news) our own Bill Shen (@williamfshen) presented PISTOL at the KDD workshop on Evaluation and Trustworthiness of Generative AI Models: genai-evaluation-kdd2024.git…. One of only a few papers selected for oral presentation 👏 arxiv.org/abs/2406.16810
1
3
10
550