English(EN)LoopCoder-v2: Only Loop Once for Efficient Test-Time Computation Scaling
新的LLM技术通过迭代改进和优化循环增强推理能力 · 跟踪5个来源
作者PulseAugur 编辑部·[10 个来源]·
研究人员开发了新的方法，通过测试时扩展来提高大型语言模型（LLM）的推理能力。REVES框架使用两阶段迭代过程来增强训练数据和优化策略，重点关注从中间的“接近但未命中”的答案中学习。另外，LoopCoder-v2探索了其Transformer架构中的最佳循环次数，发现两次循环在改进收益和计算成本之间取得了最佳平衡，更多的循环会导致收益递减。另一种方法是多序列验证器（MSV），它通过改进验证器校准来提高并行测试时扩展的效率，从而更好地选择正确的解决方案并降低延迟。
AI
arXiv:2606.19354v1 Announce Type: new Abstract: Test-time scaling (TTS) has emerged as a powerful paradigm for improving the reasoning performance of large language models (LLMs) by investing additional compute at inference time. A central component of TTS is the \emph{verifier},…
arXiv cs.CL
TIER_1English(EN)·Yuanxin Liu, Ruida Zhou, Xinyan Zhao, Amr Sharaf, Hongzhou Lin, Arijit Biswas, Mohammad Ghavamzadeh, Zhaoran Wang, Mingyi Hong·
arXiv:2606.18910v1 Announce Type: cross Abstract: Test-time scaling via sequential revision has emerged as a powerful paradigm for enhancing Large Language Model (LLM) reasoning. However, standard post-training methods primarily optimize single-shot objectives, creating a fundame…
Test-time scaling via sequential revision has emerged as a powerful paradigm for enhancing Large Language Model (LLM) reasoning. However, standard post-training methods primarily optimize single-shot objectives, creating a fundamental misalignment with multi-step inference dynami…
arXiv cs.AI
TIER_1English(EN)·Jian Yang, Shawn Guo, Wei Zhang, Tianyu Zheng, Yaxin Du, Haau-Sing Li, Jiajun Wu, Yue Song, Yan Xing, Qingsong Cai, Zelong Huang, Chuan Hao, Ran Tao, Xianglong Liu, Wayne Xin Zhao, Mingjie Tang, Weifeng Lv, Ming Zhou, Bryan Dai·
A two-stage iterative framework alternates between data augmentation and policy optimization to improve LLM reasoning by leveraging intermediate correction steps, achieving superior performance on coding benchmarks and constraint satisfaction problems.
Looped Transformers scale latent computation by repeatedly applying shared blocks, but sequential looping increases latency and KV-cache memory with the loop count. Parallel loop Transformers (PLT) alleviate this cost through cross-loop position offsets (CLP) and shared-KV gated …
arXiv cs.AI
TIER_1English(EN)·Yegon Kim, Seungyoo Lee, Chaeyun Jang, Hyungi Lee, Juho Lee·
arXiv:2603.03417v2 Announce Type: replace-cross Abstract: Parallel test-time scaling, which generates multiple candidate solutions for a single problem, is a powerful technique for improving large language model performance. However, it is hindered by two key bottlenecks: accurat…
Parallel loop Transformers achieve better code generation performance with two loops due to refined representations, while additional loops cause diminishing returns and increased positional mismatch costs.
arXiv cs.CV
TIER_1English(EN)·Hyun-Kurl Jang, Jihun Kim, Hyeokjun Kweon, Kuk-Jin Yoon·
arXiv:2606.20196v1 Announce Type: new Abstract: Continual Test-Time Adaptation (CTTA) aims to maintain model performance under evolving target domains by adapting online without labeled data. However, practical deployments often cannot retain the source dataset due to privacy or …
Continual Test-Time Adaptation (CTTA) aims to maintain model performance under evolving target domains by adapting online without labeled data. However, practical deployments often cannot retain the source dataset due to privacy or licensing constraints, and purely source-free CT…
