arXiv:2606. 00532v1 Announce Type: new Abstract: Context engineering can improve large language models without updating their weights, but mathematical reasoning exposes a key limitation: feedback accumulated in one growing prompt causes context bloat and limits the amount of learned guidance that can be used.
arXiv:2509. 25004v2 Announce Type: replace Abstract: Online reinforcement learning with verifiable rewards (RLVR) has become an effective paradigm for improving the reasoning abilities of large language models, but most methods still optimize reasoning trajectories over the static problem set, wasting rollout budget on solved or overly difficult problems.
Ladders-of-Thought (LoT) is a framework that enhances reasoning in small- to mid-scale large language models by automatically generating easier variants of reasoning problems and organizing them into difficulty buckets. It uses a self‑evolving bandit scheduler to adaptively allocate training, improving performance across math and multi‑hop reasoning tasks on 1–8 B models. LoT achieves significant gains (e.g., +32 pp on AddSub, +16 pp on QASC) and converges faster than staged curricula.
The paper introduces a verifier‑guided explainable reasoning framework for educational question answering that integrates gold‑anchored QLoRA, a task‑aware symbolic router, and group‑relative RLVR. It adapts Qwen2.5‑3B‑Instruct with field‑weighted QLoRA supervision, routes logic problems to a FOL/Z3 verifier and physics problems to a symbolic solver, and uses verifier feedback for candidate evaluation, self‑revision, and reward construction. Experiments on 438 held‑out examples show that RLVR boosts reasoning depth (P3) from 50.68 % to 72.20 %, while symbolic verification improves answer reliability at the system level.
BoundaryEvo introduces a two‑stage data‑model co‑evolution framework for distilling long‑chain reasoning from large teacher models into small language models. First, Capability Priming distills a small set of teacher trajectories to give the student an initial reasoning baseline. Then, Capability‑Boundary Co‑Evolution iteratively selects questions the student has not mastered but the teacher can solve, updating the training data as the student’s capability grows. With only 1,000 examples, BoundaryEvo improves Qwen3.5‑4B accuracy from 62.9 % to 75.3 % on IMO‑AnswerBench and from 26.8 % to 46.0 % on AMO‑Bench, outperforming static and dynamic baselines across 11 benchmarks.
arXiv:2609.00759v1 Announce Type: new Abstract: Large language models (LLMs) increasingly handle in-context learning (ICL) tasks where a long, novel context defines the rules, knowledge, and output s...
arXiv:2609.38409v1 Announce Type: new Abstract: Recent progress in large language model reasoning has been driven by benchmarks and reinforcement learning environments with automatically verifiable r...
arXiv:2606. 07720v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated remarkable reasoning abilities on mathematical and multi-hop planning tasks.
arXiv:2604.02007v3 Announce Type: replace Abstract: Building general-purpose reasoning models using reinforcement learning with verifiable rewards (RLVR) across diverse domains has been widely adopte...
arXiv:2608. 14277v1 Announce Type: cross Abstract: On-policy distillation (OPD) offers a promising way to transfer reasoning capabilities from stronger teacher models, but applying it to long-context reasoning teachers and short-context students introduces practical challenges, including tokenizer mismatch, teacher-student distribution mismatch, response length explosion, and training instability.
arXiv:2606. 09052v1 Announce Type: cross Abstract: Self-evolution offers a scalable path to stronger reasoning: a pretrained language model improves itself with only minimal external supervision.
The paper introduces masked self‑distillation, a post‑training framework that trains a language model to internalize portions of its own intermediate reasoning traces. By varying the fraction of trace internalized, the authors demonstrate that models can achieve higher inference efficiency and improved task performance on math and graph‑coloring problems. Experiments on Qwen3‑4B and Qwen3‑8B show that the method generalizes well to in‑domain out‑of‑distribution cases without catastrophic forgetting, and that supervised fine‑tuning alone can reduce trace length at the expense of generalization.
arXiv:2609.39346v1 Announce Type: new Abstract: Large language models (LLMs) offer strong reasoning capabilities but are often costly to access through commercial APIs, while small language models (S...