General learned delegation by clones

• URL: http://arxiv.org/abs/2602.13262v1
• Date: Tue, 03 Feb 2026 15:53:35 GMT
• Title: General learned delegation by clones
• Authors: Darren Li, Meiqi Chen, Chenze Shao, Fandong Meng, Jie Zhou,
• Abstract summary: Serial reasoning or uncoordinated parallel sampling can be compute-inefficient under fixed inference budgets.<n>We propose SELFCEST, which equips a base model with the ability to spawn same-weight clones in separate parallel contexts.
• Score: 55.144380092379976
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Frontier language models improve with additional test-time computation, but serial reasoning or uncoordinated parallel sampling can be compute-inefficient under fixed inference budgets. We propose SELFCEST, which equips a base model with the ability to spawn same-weight clones in separate parallel contexts by agentic reinforcement learning. Training is end-to-end under a global task reward with shared-parameter rollouts, yielding a learned controller that allocates both generation and context budget across branches. Across challenging math reasoning benchmarks and long-context multi-hop QA, SELFCEST improves the accuracy-cost Pareto frontier relative to monolithic baselines at matched inference budget, and exhibits out-of-distribution generalization in both domains.

Related papers

• $V_1$: Unifying Generation and Self-Verification for Parallel Reasoners [69.66089681814013]
  $V_$ is a framework that unifies generation and verification through efficient pairwise ranking.<n>$V_$-Infer improves Pass@1 by up to $10%$ over pointwise verification.<n>$V_$-PairRL achieves $7$--$9%$ test-time scaling gains over standard RL and pointwise joint training.
  arXiv  Detail & Related papers  (2026-03-04T17:22:16Z)
• CoBA-RL: Capability-Oriented Budget Allocation for Reinforcement Learning in LLMs [31.371566320424552]
  CoBA-RL is a reinforcement learning algorithm designed to adaptively allocate rollout budgets based on the model's evolving capability.<n>Our approach effectively orchestrates the trade-off between exploration and exploitation, delivering consistent generalization improvements.
  arXiv  Detail & Related papers  (2026-02-03T03:14:36Z)
• Towards regularized learning from functional data with covariate shift [3.072411352294816]
  This paper investigates a general regularization framework for unsupervised domain adaptation in vector-valued regression.<n>By restricting the hypothesis space, we develop a practical operator learning algorithm capable of handling functional outputs.
  arXiv  Detail & Related papers  (2026-01-28T20:30:05Z)
• Coupled Variational Reinforcement Learning for Language Model General Reasoning [83.82392089177841]
  We propose textitbCoupled bVari bReinforcement bLearning (CoVRL) to bridge variational inference and reinforcement learning.<n>CoVRL improves performance by 12.4% over the base model and achieves an additional 2.3% improvement over strong state-of-the-art verifier-free RL baselines.
  arXiv  Detail & Related papers  (2025-12-14T07:03:51Z)
• ThreadWeaver: Adaptive Threading for Efficient Parallel Reasoning in Language Models [99.6720868215076]
  We introduce ThreadWeaver, a framework for adaptive parallel reasoning.<n> ThreadWeaver achieves accuracy on par with popular sequential reasoning models of comparable size.<n>We show that ThreadWeaver delivers up to 1.53x average speedup in token latency.
  arXiv  Detail & Related papers  (2025-11-24T18:55:59Z)
• Context Attribution with Multi-Armed Bandit Optimization [11.715006981206844]
  We propose a novel framework that formulates context attribution as a multi-armed bandit (CMAB) problem.<n>We employ Combinatorial Thompson Sampling (CTS) to efficiently explore the exponentially large space of context subsets under a limited query budget.<n>Our method defines a reward function based on normalized token likelihoods, capturing how well a subset of segments supports the original model response.
  arXiv  Detail & Related papers  (2025-06-24T19:47:27Z)
• DynScaling: Efficient Verifier-free Inference Scaling via Dynamic and Integrated Sampling [20.605487145370752]
  Inference-time scaling has proven effective in boosting large language model (LLM) performance through increased test-time computation.<n>Yet, its practical application is often hindered by reliance on external verifiers or a lack of optimization for realistic computational constraints.<n>We propose DynScaling, which addresses these limitations through two primary innovations: an integrated parallel-sequential sampling strategy and a bandit-based dynamic budget allocation framework.
  arXiv  Detail & Related papers  (2025-06-19T05:40:54Z)
• CC-LEARN: Cohort-based Consistency Learning [5.7716971260066]
  Large language models struggle with consistent, robust reasoning.<n>We introduce cohort-based Consistency Learning (CC-Learn)<n>Experiments show that CC-Learn boosts both accuracy and reasoning stability over pretrained and SFT baselines.
  arXiv  Detail & Related papers  (2025-06-18T17:41:28Z)
• SPARE: Single-Pass Annotation with Reference-Guided Evaluation for Automatic Process Supervision and Reward Modelling [58.05959902776133]
  We introduce Single-Pass.<n>with Reference-Guided Evaluation (SPARE), a novel structured framework that enables efficient per-step annotation.<n>We demonstrate SPARE's effectiveness across four diverse datasets spanning mathematical reasoning (GSM8K, MATH), multi-hop question answering (MuSiQue-Ans), and spatial reasoning (SpaRP)<n>On ProcessBench, SPARE demonstrates data-efficient out-of-distribution generalization, using only $sim$16% of training samples compared to human-labeled and other synthetically trained baselines.
  arXiv  Detail & Related papers  (2025-06-18T14:37:59Z)
• Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning [55.15106182268834]
  Reinforcement learning with verifiable rewards (RLVR) has emerged as the leading approach for enhancing reasoning capabilities in large language models.<n>It faces a fundamental compute and memory asymmetry: rollout generation is embarrassingly parallel and memory-light, whereas policy updates are communication-heavy and memory-intensive.<n>We introduce PODS (Policy Optimization with Down-Sampling), which decouples rollout generation from policy updates by training only on a strategically selected subset of rollouts.
  arXiv  Detail & Related papers  (2025-04-18T17:49:55Z)
• A Unified Framework for Multi-distribution Density Ratio Estimation [101.67420298343512]
  Binary density ratio estimation (DRE) provides the foundation for many state-of-the-art machine learning algorithms. We develop a general framework from the perspective of Bregman minimization divergence. We show that our framework leads to methods that strictly generalize their counterparts in binary DRE.
  arXiv  Detail & Related papers  (2021-12-07T01:23:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.