目录
  • SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents
  • ReMA: Learning to Meta-think for LLMs with Multi-Agent Reinforcement Learning
  • Risk-Sensitive RL for Alleviating Exploration Dilemmas in Large Language Models
  • PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier
  • Learning to Learn with Contrastive Meta-Objective
  • Multi-Type Preference Learning: Empowering Preference-Based Reinforcement Learning with Equal Preferences
  • MetaCURE: Meta Reinforcement Learning with Empowerment-Driven Exploration
  • Absolute Zero: Reinforced Self-play Reasoning with Zero Data
  • auto-curriculum learning (Jiang et al., 2021b)
  • Meta-Motivo(Tirinzoni 等人,2025),zero-shot goal-conditioned RL
  • Unsupervised Skill Discovery via Recurrent Skill Training
  • Learning to Discover Skills through Guidance
  • One After Another: Learning Incremental Skills for a Changing World
  • Direct then Diffuse: Incremental Unsupervised Skill Discovery for State Covering and Goal Reaching
  • Horizon Generalization in Reinforcement Learning
  • HIQL: Offline Goal-Conditioned RL with Latent States as Actions
  • Contrastive Preference Learning: Learning from Human Feedback without RL
  • Few is More: Task-Efficient Skill-Discovery for Multi-Task Offline Multi-Agent Reinforcement Learning
  • Rethinking Reward Modeling in Preference-based Large Language Model Alignment
  • DOPL: Direct Online Preference Learning for Restless Bandits with Preference Feedback
  • Data Center Cooling System Optimization Using Offline Reinforcement Learning
  • SpikeLLM: Scaling up Spiking Neural Network to Large Language Models via Saliency-based Spiking
  • Rethinking Inverse Reinforcement Learning: from Data Alignment to Task Alignment
  • Why Distillation can Outperform Zero-RL: The Role of Flexible Reasoning
  • Thinkless: LLM Learns When to Think
  • Learning to Reason without External Rewards


SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents

  • arXiv:https://arxiv.org/abs/2609.08149
  • PDF:https://arxiv.org/pdf/2609.08149
  • HTML:https://arxiv.org/html/2609.08149
  • GitHub:https://github.com/open-compass/AgentCompass
  • Dataset:https://huggingface.co/datasets/opencompass/SWEBench-Pro-Verifie
  • evaluation guide:https://agent-compass.mintlify.app/en/user_guide/modules/benchmarks/swebench_pro_verified
  • 来源:最近关心的方向。
  • 内容概括(还没有读):研究 SWE-bench Pro 中的两类可靠性问题:gold solution / hidden evaluation 泄露导致的 reward hacking,以及题目描述和测试质量问题。通过 anti-hacking safeguards 和 task refinement 构建更可信的 Benchmark。

ReMA: Learning to Meta-think for LLMs with Multi-Agent Reinforcement Learning

  • 来源:曾经想看的正样本文章。
  • arxiv:https://arxiv.org/abs/2503.09501
  • pdf:https://arxiv.org/pdf/2503.09501
  • html:https://arxiv.org/html/2503.09501v3
  • GitHub:https://github.com/ziyuwan/ReMA-public

Risk-Sensitive RL for Alleviating Exploration Dilemmas in Large Language Models

  • 来源:曾经想看的正样本文章。
  • arxiv:https://arxiv.org/abs/2509.24261
  • pdf:https://arxiv.org/pdf/2509.24261
  • html:https://arxiv.org/html/2509.24261v1
  • GitHub:https://github.com/Jackory/RS-GRPO

感觉 intro 和 related work 写的真好啊。

这个当时简单看了一下。

PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier

  • 来源:曾经想看的正样本文章。
  • arxiv:https://arxiv.org/abs/2506.10406
  • pdf:https://arxiv.org/pdf/2506.10406
  • html 坏掉了。
  • website:https://jackory.github.io/pag/
  • GitHub:https://github.com/Jackory/Policy-As-GenVerifier

Learning to Learn with Contrastive Meta-Objective

  • 来源:无意中看到的,NeurIPS 2025 oral。
  • arxiv:https://arxiv.org/abs/2410.05975

(还没读。这篇文章看起来比较古典,做的是传统 ML,并不是做 llm 的。 (这个东西能用在 llm 上吗?现在看到一个东西,就会想它能否用在 llm 上

Multi-Type Preference Learning: Empowering Preference-Based Reinforcement Learning with Equal Preferences

  • 来源:无意中搜到的。ICRA 2025。
  • arxiv:https://arxiv.org/abs/2409.07268
  • GitHub:https://github.com/FeiCuiLengMMbb/paper_MTPL
  • 好奇是不是 multi-type + PbRL。

MetaCURE: Meta Reinforcement Learning with Empowerment-Driven Exploration

  • arxiv:https://arxiv.org/abs/2006.08170
  • 来源:合作者说有趣的 skill + meta-RL 论文,ICML 2021。

Absolute Zero: Reinforced Self-play Reasoning with Zero Data

  • arxiv:https://arxiv.org/abs/2505.03335
  • 来源:neurips 2025 best paper 的一作 yue yang 的 NeurIPS 2025 spotlight 工作。被题目吸引住了,单纯好奇,想读一读。

auto-curriculum learning (Jiang et al., 2021b)

  • 来源:RSD。似乎可以做自动 curriculum learning,或许是有启发性的。

Meta-Motivo(Tirinzoni 等人,2025),zero-shot goal-conditioned RL

  • 来源:RGSD。可能包含一个技能库,也想看。速读一下就行。

Unsupervised Skill Discovery via Recurrent Skill Training

  • 来源:合作者推荐的 skill discovery 先前工作。

Learning to Discover Skills through Guidance

  • 来源:同上。

One After Another: Learning Incremental Skills for a Changing World

  • 来源:同上。

Direct then Diffuse: Incremental Unsupervised Skill Discovery for State Covering and Goal Reaching

  • 来源:同上。

Horizon Generalization in Reinforcement Learning

  • arxiv:https://arxiv.org/abs/2501.02709
  • website:https://horizon-generalization.github.io/
  • 来源:Benjamin Eysenbach 的新作,是一篇 arxiv paper,同学说有趣。

HIQL: Offline Goal-Conditioned RL with Latent States as Actions

  • arxiv:https://arxiv.org/abs/2307.11949
  • website:https://seohong.me/projects/hiql/
  • 来源:合作者推荐的文章,好像也是 Benjamin Eysenbach 发表的。

Contrastive Preference Learning: Learning from Human Feedback without RL

  • arxiv:https://arxiv.org/abs/2310.13639
  • GitHub:https://github.com/jhejna/cpl
  • 来源:无意中搜到的文章,ICLR 2024,好像之前读过。
  • 主要内容:

Few is More: Task-Efficient Skill-Discovery for Multi-Task Offline Multi-Agent Reinforcement Learning

  • arxiv:https://arxiv.org/abs/2502.08985
  • 来源:同学的最新工作。
  • 主要内容:
    • 这篇文章关注的 setting 是 offline multi-task MARL;特别的,agent 只在(比如说)三个人合作的场景上训练,然后就可以泛化到任意多个人合作的场景。同学讲的故事是,用 transformer 作为一个翻译器,把三个人的合作动作翻译为多个人的,感觉这个故事听起来非常好。

Rethinking Reward Modeling in Preference-based Large Language Model Alignment

  • arxiv:https://arxiv.org/abs/2411.04991
  • OpenReview:https://openreview.net/forum?id=rfdblE10qm
  • 来源:ICLR 2025 oral。
  • 主要内容:
    • 这篇文章关注 LLM 的 RLHF。据说不采用 bradley-terry model 来建模 reward model,而是直接训一个分类器,学习一个 (x,y) 是好的还剩坏的,然后使用分类器的概率 logit 作为 RLHF 的 reward。
    • 是否使用了非成对的比较 \((x_1, y_1^+, x_2, y_2^-)\),而非把成对比较 \((x, y^+, y^-)\) 打乱(?)
    • 实验是否过于 toy(?)理论大概说了什么(?)

DOPL: Direct Online Preference Learning for Restless Bandits with Preference Feedback

  • arxiv:https://arxiv.org/abs/2410.05527
  • open review:https://openreview.net/forum?id=2iYVBqRHK4
  • 来源:合作者推荐的文章。
  • 主要内容:
    • preference-based index policy(?)
  • whittle index,一个结论,两个等价条件,经典问题的证明方式。

Data Center Cooling System Optimization Using Offline Reinforcement Learning

  • arxiv:https://arxiv.org/pdf/2501.15085
  • 来源:xianyuan zhan 组的新文章。
  • 主要内容:
    • T-symmetry。

SpikeLLM: Scaling up Spiking Neural Network to Large Language Models via Saliency-based Spiking

  • arxiv:https://arxiv.org/abs/2407.04752
  • 来源:师兄推荐的神秘文章,ICLR 2025 poster。

Rethinking Inverse Reinforcement Learning: from Data Alignment to Task Alignment

  • arxiv:https://arxiv.org/abs/2410.23680
  • 来源:偶然看到的文章。

Why Distillation can Outperform Zero-RL: The Role of Flexible Reasoning

  • arxiv:https://arxiv.org/abs/2505.21067
  • 来源:偶然看到的文章。

Thinkless: LLM Learns When to Think

  • arxiv:https://arxiv.org/abs/2505.13379
  • 来源:偶然看到的文章。

Learning to Reason without External Rewards

  • arxiv:https://arxiv.org/abs/2505.19590
  • 来源:偶然看到的文章。