Metraipolis › archive › dc_7143fa8bea65
FADE: Factored Ability--Difficulty Filtering for Online RLVR Scheduling — arXiv:2611.03625
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.03625
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:23:25.199Z
- SHA-256 of the archived text
548c0959ab91dbd10d0d08a68239aeddfd85b95b6561fa1de6b7e29eebb379ae- Stable URL of this page
https://metraipolis.com/documento/dc_7143fa8bea65
How to cite
FADE: Factored Ability--Difficulty Filtering for Online RLVR Scheduling — arXiv:2611.03625 — arXiv (en), https://arxiv.org/abs/2611.03625 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7143fa8bea65 (Metraipolis archive dc_7143fa8bea65, sha256 548c0959ab91dbd10d0d08a68239aeddfd85b95b6561fa1de6b7e29eebb379ae)
BibTeX
@misc{dc7143fa8bea65,
title = {FADE: Factored Ability--Difficulty Filtering for Online RLVR Scheduling — arXiv:2611.03625},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.03625},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7143fa8bea65; sha256 548c0959ab91dbd10d0d08a68239aeddfd85b95b6561fa1de6b7e29eebb379ae},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.03625. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Group-based reinforcement learning with verifiable rewards (RLVR) learns only from prompts whose sampled completions disagree: all-pass and all-fail groups produce zero relative advantage and burn generation budget without moving the policy. Adaptive schedulers try to steer rollouts toward informative prompts, yet they rely on per-prompt pass-rate trackers that entangle two quantities evolving at very different speeds---the policy's shifting competence across reasoning skills and each prompt's intrinsic difficulty. When a single exponential average or Beta counter tracks both at once, short memory scrambles the relative ordering of prompts whereas long memory reports stale pass rates. We introduce FADE (Factored Ability--Difficulty Estimation), a bilinear state-space filter that separates the two timescales. FADE represents each prompt's success logit as the inner product of a whitened $K$-dimensional skill loading with a fast-drifting policy ability vector, minus a slow-shrinking scalar item intercept. Using Pólya--Gamma augmentation, each minibatch of binary rollout outcomes updates the $K$-dimensional ability posterior in closed form via a rank-$K$ Woodbury step, instantly transferring competence shifts to all unobserved prompts that share those skills while leaving item intercepts stable. From the resulting posterior mean and variance we derive a closed-form utility for the probability of drawing a non-degenerate rollout group. Across five reasoning benchmarks, two base models, and four scheduling paradigms under a matched rollout budget, FADE consistently improves downstream accuracy and calibration while adding less than $0.6$\,ms per step. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.03625 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-03T11:02:37.731Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).