11 papers
Muse Glimmer 30B: An On-Device Agentic Model
Arash Ebrahim Abtahizadeh, Muna Aghamelu, Adil Ahmed +327
We introduce Muse Glimmer, an open-source 30B parameter causal language model with perception encoder, purpose-built for autonomous agentic tasks on consumer hardware. Muse Glimmer…
AI Research Preference Models
Thomas Simon Foster, Bassel Al Omari, Tingchen Fu +30
AI research agents (AIRA) can now carry machine learning experiments from proposal through implementation and evaluation. Yet progress on frontier tasks is throttled by the cost of…
Training AI Scientists to Replicate Research
Damon Falck, Samer Sabri, Anja Surina +8
The replicability of papers is a cornerstone of scientific knowledge, ensuring the reliability of existing results and providing a base for further experiments. The act of replicat…
CODA: Coordination via On-Policy Diffusion for Multi-Agent Offline Reinforcement Learning
Marcel Hedman, Kale-ab Abebe Tessera, Juan Claude Formanek +5
Offline multi-agent reinforcement learning (MARL) enables policy learning from fixed datasets, but is prone to coordination failure: agents trained on static, off-policy data conve…
Stochasticity in Tokenisation Improves Robustness
Sophie Steger, Rui Li, Sofiane Ennadir +4
The widespread adoption of large language models (LLMs) has increased concerns about their robustness. Vulnerabilities in perturbations of tokenisation of the input indicate that m…
Evolution Strategies at the Hyperscale
Bidipta Sarkar, Mattie Fellows, Juan Agustin Duque +17
Evolution Strategies (ES) is a class of powerful black-box optimisation methods that are highly parallelisable and can handle non-differentiable and noisy objectives. However, naïv…
GEM: A Gym for Agentic LLMs
Zichen Liu, Anya Sims, Keyu Duan +16
The training paradigm for large language models (LLMs) is moving from static datasets to experience-based learning, where agents acquire skills via interacting with complex environ…
Deep Thinking by Markov Chain of Continuous Thoughts
Jiayu Liu, Zhenya Huang, Xuan Yang +6
Transformer-based models can perform complicated reasoning by generating reasoning paths token by token. While effective, this approach often requires generating thousands of token…
StochasTok: Improving Fine-Grained Subword Understanding in LLMs
Anya Sims, Thom Foster, Klara Kaleb +5
Subword-level understanding is integral to numerous tasks, including understanding multi-digit numbers, spelling mistakes, abbreviations, rhyming, and wordplay. Despite this, curre…
Reinforcing General Reasoning without Verifiers
Xiangxin Zhou, Zichen Liu, Anya Sims +6
The recent paradigm shift towards training large language models (LLMs) using DeepSeek-R1-Zero-style reinforcement learning (RL) on verifiable rewards has led to impressive advance…
The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning
Anya Sims, Cong Lu, Jakob Foerster +1
Offline reinforcement learning aims to train agents from pre-collected datasets. However, this comes with the added challenge of estimating the value of behaviors not covered in th…