Maven is an RL method using answer-conditioned evidence-state values to assign rewards to add, link, and drop actions on evidence memory, outperforming outcome-only baselines on LongBench v2, LongReason, and RULER.
arXiv preprint arXiv:2602.05758 , year =
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 5verdicts
UNVERDICTED 5roles
background 1polarities
unclear 1representative citing papers
CVT-RL improves verified task success to 78.9% and reduces hacking to 3.9% in long-horizon language agents by combining intervention-validity gating with a selection-adjusted doubly robust PCCC estimator.
SP-KV trains a utility predictor jointly with the LLM to dynamically prune low-utility KV cache entries, achieving 3-10x memory reduction during generation with negligible performance loss.
RTT bridges response-level rubrics to token-level rewards via a relevance discriminator and intra-sample group normalization, yielding higher instruction and rubric accuracy than baselines.
LongAct uses saliency from high-magnitude activations to guide sparse weight updates in long-context RL, yielding about 8% gains on LongBench v2 across multiple algorithms.
citing papers explorer
-
Evidence-State Rewards for Long-Context Reasoning
Maven is an RL method using answer-conditioned evidence-state values to assign rewards to add, link, and drop actions on evidence memory, outperforming outcome-only baselines on LongBench v2, LongReason, and RULER.