Learning to act in unfamiliar environments requires agents to infer how the world works and revise that understanding as new evidence arrives. Yet limited observations can support multiple world models that explain past interactions but predict different outcomes in unseen states. We introduce Memento 3, building on th...
Hao-Yu Zhao, Zhengxu Yu, Zhi-Yuan He et al.· 3 citations
Distilling the reasoning capabilities of large language models (LLMs) into smaller students is a central challenge for efficient deployment. Current approaches face a fundamental tension: optimizing purely for verifiable task rewards (e.g., via GRPO) leads to reward hacking, where students arrive at correct final answe...
Matthieu Zimmer, Xiao-Tong Ji, Tu Nguyen et al.· 0 citations
This work traces the retry loop to the adaptivity of masked decoding, and analyses an invariance proposition: the task-blind factor cancels exactly from the reverse conditional, which coincides with the task posterior of an idealized uncorrupted model.
Jia-Cheng Qiu, Christopher E. Mower, Jan Peters et al.· arXiv.org· 0 citations
A tail influence for each queryable conditional law that aggregates how its uncertainty affects CVaR across every Bellman reuse is derived, and an exact-grid regime in which tail- and mean-optimal allocations coincide is characterized.
P. Bourigault, Xiao-Tong Ji, Matthieu Zimmer et al.· 1 citation
Test-time self-distillation is proposed, a decoding-time method that extracts a steering signal from the self-distillation framework without any parameter updates, reward models, or training data, demonstrating that the self-distillation principle can be operationalized at inference time.
S. Tan, Xiao-Tong Ji, Rasul Tutunov et al.· arXiv.org· 0 citations
CompoSimplex is introduced, a library with configurable support rules, regularisation primitives, and simplex solvers for constructing and evaluating compositional decoders, and it is shown that compositions can realise trade-offs between single-sample quality, multi-sample quality, and diversity that are not attained...
Xiao-Tong Ji, Ahmed Khaled Khamis, Rasul Tutunov et al.· 0 citations
Controling how much and when teacher information is introduced improves specialisation while preserving broader capability is shown, and results show that controlling how much and when teacher information is introduced improves specialisation while preserving broader capability.
Ahmed Khaled Khamis, Xiao-Tong Ji, Hassan Jaber et al.· arXiv.org· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.