bot: update RSS feed

2025-10-12 00:02:36 +00:00
parent 832ab9c702
commit 30a6a210db
1 changed files with 31 additions and 31 deletions
--- a/feed.xml
+++ b/feed.xml
@@ -7,32 +7,32 @@
    <docs>http://www.rssboard.org/rss-specification</docs>
    <generator>python-feedgen</generator>
    <language>en</language>
-    <lastBuildDate>Sat, 11 Oct 2025 00:02:28 +0000</lastBuildDate>
+    <lastBuildDate>Sun, 12 Oct 2025 00:02:36 +0000</lastBuildDate>
    <item>
      <title>Meta-Awareness Enhances Reasoning Models: Self-Alignment Reinforcement   Learning</title>
      <link>https://arxiv.org/abs/2510.03259</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03259.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yoonjeon Kim, Doohyuk Jang, Eunho Yang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 38&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent studies on reasoning models explore the meta-awareness of language models, the ability to know how to think by itself. We argue that large reasoning models lack this meta-awareness property by proving severe misalignment between true rollouts and predicted meta information. We posit that aligning meta-prediction with true rollouts will lead to significant performance gains. To verify this hypothesis, we design a training pipeline that boosts Meta-Awareness via Self-Alignment (MASA), and prove that enhanced meta-awareness directly translates to improved accuracy. Unlike existing meta-cognitive reasoning models, our method does not require external training sources but leverages self-generated signals to train meta-awareness. Moreover, our method enables efficient training by i) filtering out zero-variance prompts that are either trivial or unsolvable and ii) cutting off lengthy rollouts when they are unlikely to lead to correct answers. The results are inspiring: our strategy yields significant improvements in both accuracy and training efficiency on in-domain tasks and shows strong generalization to out-of-domain benchmarks. More specifically, our method can speed up GRPO training by over 1.28x to reach the same performance, and achieve a 19.3% gain in accuracy on AIME25, and a 6.2 % average gain over six mathematics benchmarks. Training with meta-cognitive guidance enhances out-of-domain generalization, giving a 3.87 % boost on GPQA-Diamond and a 2.08 % overall accuracy gain across 13 benchmarks spanning logical, scientific, and coding domains.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03259.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yoonjeon Kim, Doohyuk Jang, Eunho Yang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 42&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent studies on reasoning models explore the meta-awareness of language models, the ability to know how to think by itself. We argue that large reasoning models lack this meta-awareness property by proving severe misalignment between true rollouts and predicted meta information. We posit that aligning meta-prediction with true rollouts will lead to significant performance gains. To verify this hypothesis, we design a training pipeline that boosts Meta-Awareness via Self-Alignment (MASA), and prove that enhanced meta-awareness directly translates to improved accuracy. Unlike existing meta-cognitive reasoning models, our method does not require external training sources but leverages self-generated signals to train meta-awareness. Moreover, our method enables efficient training by i) filtering out zero-variance prompts that are either trivial or unsolvable and ii) cutting off lengthy rollouts when they are unlikely to lead to correct answers. The results are inspiring: our strategy yields significant improvements in both accuracy and training efficiency on in-domain tasks and shows strong generalization to out-of-domain benchmarks. More specifically, our method can speed up GRPO training by over 1.28x to reach the same performance, and achieve a 19.3% gain in accuracy on AIME25, and a 6.2 % average gain over six mathematics benchmarks. Training with meta-cognitive guidance enhances out-of-domain generalization, giving a 3.87 % boost on GPQA-Diamond and a 2.08 % overall accuracy gain across 13 benchmarks spanning logical, scientific, and coding domains.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.03259</guid>
      <pubDate>Fri, 26 Sep 2025 14:05:48 +0000</pubDate>
    </item>
    <item>
      <title>Beyond Outliers: A Study of Optimizers Under Quantization</title>
      <link>https://arxiv.org/abs/2509.23500</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.23500.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Georgios Vlassis, Saleh Ashkboos, Alexandra Volkova, Torsten Hoefler, Dan Alistarh&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; As new optimizers gain traction and model quantization becomes standard for efficient deployment, a key question arises: how does the choice of optimizer affect model performance in the presence of quantization? Despite progress in both areas, systematic evidence on optimizer-quantization interactions remains limited. To fill this gap, we study the impact of optimizer choice on model robustness under quantization, considering both post-training quantization (PTQ), and quantization-aware training (QAT). We first train full-precision models, ranging from 50M to 1.5B parameters, with six optimizers, to explore the hyperparameter landscape, and establish well-tuned baselines. We then apply PTQ to evaluate how model performance degrades when trained with different optimizers. We find that outlier-related metrics, such as the max-to-mean ratio (MMR) and Kurtosis, fail to predict the PTQ performance across different optimizers. We show analytically that this is due to the MMR capturing only isolated layer errors, while ignoring how quantization errors accumulate and propagate through the network. To study the QAT degradation, we train quantized models from scratch and compare them to our original-precision baselines. We find that optimizers performing well in the original pretraining setup may not remain optimal under QAT, and that models trained with Shampoo show the lowest accuracy degradation. Finally, we derive scaling laws for quantization-aware training under different optimizers, showing that Shampoo achieves the highest parameter efficiency of all tested optimizers.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.23500.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Georgios Vlassis, Saleh Ashkboos, Alexandra Volkova, Torsten Hoefler, Dan Alistarh&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; As new optimizers gain traction and model quantization becomes standard for efficient deployment, a key question arises: how does the choice of optimizer affect model performance in the presence of quantization? Despite progress in both areas, systematic evidence on optimizer-quantization interactions remains limited. To fill this gap, we study the impact of optimizer choice on model robustness under quantization, considering both post-training quantization (PTQ), and quantization-aware training (QAT). We first train full-precision models, ranging from 50M to 1.5B parameters, with six optimizers, to explore the hyperparameter landscape, and establish well-tuned baselines. We then apply PTQ to evaluate how model performance degrades when trained with different optimizers. We find that outlier-related metrics, such as the max-to-mean ratio (MMR) and Kurtosis, fail to predict the PTQ performance across different optimizers. We show analytically that this is due to the MMR capturing only isolated layer errors, while ignoring how quantization errors accumulate and propagate through the network. To study the QAT degradation, we train quantized models from scratch and compare them to our original-precision baselines. We find that optimizers performing well in the original pretraining setup may not remain optimal under QAT, and that models trained with Shampoo show the lowest accuracy degradation. Finally, we derive scaling laws for quantization-aware training under different optimizers, showing that Shampoo achieves the highest parameter efficiency of all tested optimizers.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2509.23500</guid>
      <pubDate>Sat, 27 Sep 2025 21:15:22 +0000</pubDate>
    </item>
    <item>
      <title>From What to Why: A Multi-Agent System for Evidence-based Chemical   Reaction Condition Reasoning</title>
      <link>https://arxiv.org/abs/2509.23768</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.23768.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Cheng Yang, Jiaxuan Lu, Haiyuan Wan, Junchi Yu, Feiwei Qin&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 42&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; The chemical reaction recommendation is to select proper reaction condition parameters for chemical reactions, which is pivotal to accelerating chemical science. With the rapid development of large language models (LLMs), there is growing interest in leveraging their reasoning and planning capabilities for reaction condition recommendation. Despite their success, existing methods rarely explain the rationale behind the recommended reaction conditions, limiting their utility in high-stakes scientific workflows. In this work, we propose ChemMAS, a multi-agent system that reframes condition prediction as an evidence-based reasoning task. ChemMAS decomposes the task into mechanistic grounding, multi-channel recall, constraint-aware agentic debate, and rationale aggregation. Each decision is backed by interpretable justifications grounded in chemical knowledge and retrieved precedents. Experiments show that ChemMAS achieves 20-35% gains over domain-specific baselines and outperforms general-purpose LLMs by 10-15% in Top-1 accuracy, while offering falsifiable, human-trustable rationales, which establishes a new paradigm for explainable AI in scientific discovery.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.23768.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Cheng Yang, Jiaxuan Lu, Haiyuan Wan, Junchi Yu, Feiwei Qin&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 45&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; The chemical reaction recommendation is to select proper reaction condition parameters for chemical reactions, which is pivotal to accelerating chemical science. With the rapid development of large language models (LLMs), there is growing interest in leveraging their reasoning and planning capabilities for reaction condition recommendation. Despite their success, existing methods rarely explain the rationale behind the recommended reaction conditions, limiting their utility in high-stakes scientific workflows. In this work, we propose ChemMAS, a multi-agent system that reframes condition prediction as an evidence-based reasoning task. ChemMAS decomposes the task into mechanistic grounding, multi-channel recall, constraint-aware agentic debate, and rationale aggregation. Each decision is backed by interpretable justifications grounded in chemical knowledge and retrieved precedents. Experiments show that ChemMAS achieves 20-35% gains over domain-specific baselines and outperforms general-purpose LLMs by 10-15% in Top-1 accuracy, while offering falsifiable, human-trustable rationales, which establishes a new paradigm for explainable AI in scientific discovery.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2509.23768</guid>
      <pubDate>Sun, 28 Sep 2025 09:34:35 +0000</pubDate>
    </item>
    <item>
      <title>MemMamba: Rethinking Memory Patterns in State Space Model</title>
      <link>https://arxiv.org/abs/2510.03279</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03279.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Youjin Wang, Yangjingyi Chen, Jiahao Yan, Jiaxuan Lu, Xiao Sun&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 55&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; With the explosive growth of data, long-sequence modeling has become increasingly important in tasks such as natural language processing and bioinformatics. However, existing methods face inherent trade-offs between efficiency and memory. Recurrent neural networks suffer from gradient vanishing and explosion, making them hard to scale. Transformers can model global dependencies but are constrained by quadratic complexity. Recently, selective state-space models such as Mamba have demonstrated high efficiency with O(n) time and O(1) recurrent inference, yet their long-range memory decays exponentially. In this work, we conduct mathematical derivations and information-theoretic analysis to systematically uncover the memory decay mechanism of Mamba, answering a fundamental question: what is the nature of Mamba's long-range memory and how does it retain information? To quantify key information loss, we further introduce horizontal-vertical memory fidelity metrics that capture degradation both within and across layers. Inspired by how humans distill and retain salient information when reading long documents, we propose MemMamba, a novel architectural framework that integrates state summarization mechanism together with cross-layer and cross-token attention, which alleviates long-range forgetting while preserving linear complexity. MemMamba achieves significant improvements over existing Mamba variants and Transformers on long-sequence benchmarks such as PG19 and Passkey Retrieval, while delivering a 48% speedup in inference efficiency. Both theoretical analysis and empirical results demonstrate that MemMamba achieves a breakthrough in the complexity-memory trade-off, offering a new paradigm for ultra-long sequence modeling.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03279.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Youjin Wang, Yangjingyi Chen, Jiahao Yan, Jiaxuan Lu, Xiao Sun&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 63&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; With the explosive growth of data, long-sequence modeling has become increasingly important in tasks such as natural language processing and bioinformatics. However, existing methods face inherent trade-offs between efficiency and memory. Recurrent neural networks suffer from gradient vanishing and explosion, making them hard to scale. Transformers can model global dependencies but are constrained by quadratic complexity. Recently, selective state-space models such as Mamba have demonstrated high efficiency with O(n) time and O(1) recurrent inference, yet their long-range memory decays exponentially. In this work, we conduct mathematical derivations and information-theoretic analysis to systematically uncover the memory decay mechanism of Mamba, answering a fundamental question: what is the nature of Mamba's long-range memory and how does it retain information? To quantify key information loss, we further introduce horizontal-vertical memory fidelity metrics that capture degradation both within and across layers. Inspired by how humans distill and retain salient information when reading long documents, we propose MemMamba, a novel architectural framework that integrates state summarization mechanism together with cross-layer and cross-token attention, which alleviates long-range forgetting while preserving linear complexity. MemMamba achieves significant improvements over existing Mamba variants and Transformers on long-sequence benchmarks such as PG19 and Passkey Retrieval, while delivering a 48% speedup in inference efficiency. Both theoretical analysis and empirical results demonstrate that MemMamba achieves a breakthrough in the complexity-memory trade-off, offering a new paradigm for ultra-long sequence modeling.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.03279</guid>
      <pubDate>Sun, 28 Sep 2025 14:40:58 +0000</pubDate>
    </item>
@@ -46,14 +46,14 @@
    <item>
      <title>UP2You: Fast Reconstruction of Yourself from Unconstrained Photo   Collections</title>
      <link>https://arxiv.org/abs/2509.24817</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.24817.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Zeyu Cai, Ziyang Li, Xiaoben Li, Boqian Li, Zeyu Wang, Zhenyu Zhang, Yuliang Xiu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 3&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We present UP2You, the first tuning-free solution for reconstructing high-fidelity 3D clothed portraits from extremely unconstrained in-the-wild 2D photos. Unlike previous approaches that require "clean" inputs (e.g., full-body images with minimal occlusions, or well-calibrated cross-view captures), UP2You directly processes raw, unstructured photographs, which may vary significantly in pose, viewpoint, cropping, and occlusion. Instead of compressing data into tokens for slow online text-to-3D optimization, we introduce a data rectifier paradigm that efficiently converts unconstrained inputs into clean, orthogonal multi-view images in a single forward pass within seconds, simplifying the 3D reconstruction. Central to UP2You is a pose-correlated feature aggregation module (PCFA), that selectively fuses information from multiple reference images w.r.t. target poses, enabling better identity preservation and nearly constant memory footprint, with more observations. We also introduce a perceiver-based multi-reference shape predictor, removing the need for pre-captured body templates. Extensive experiments on 4D-Dress, PuzzleIOI, and in-the-wild captures demonstrate that UP2You consistently surpasses previous methods in both geometric accuracy (Chamfer-15%, P2S-18% on PuzzleIOI) and texture fidelity (PSNR-21%, LPIPS-46% on 4D-Dress). UP2You is efficient (1.5 minutes per person), and versatile (supports arbitrary pose control, and training-free multi-garment 3D virtual try-on), making it practical for real-world scenarios where humans are casually captured. Both models and code will be released to facilitate future research on this underexplored task. Project Page: https://zcai0612.github.io/UP2You&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.24817.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Zeyu Cai, Ziyang Li, Xiaoben Li, Boqian Li, Zeyu Wang, Zhenyu Zhang, Yuliang Xiu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 5&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We present UP2You, the first tuning-free solution for reconstructing high-fidelity 3D clothed portraits from extremely unconstrained in-the-wild 2D photos. Unlike previous approaches that require "clean" inputs (e.g., full-body images with minimal occlusions, or well-calibrated cross-view captures), UP2You directly processes raw, unstructured photographs, which may vary significantly in pose, viewpoint, cropping, and occlusion. Instead of compressing data into tokens for slow online text-to-3D optimization, we introduce a data rectifier paradigm that efficiently converts unconstrained inputs into clean, orthogonal multi-view images in a single forward pass within seconds, simplifying the 3D reconstruction. Central to UP2You is a pose-correlated feature aggregation module (PCFA), that selectively fuses information from multiple reference images w.r.t. target poses, enabling better identity preservation and nearly constant memory footprint, with more observations. We also introduce a perceiver-based multi-reference shape predictor, removing the need for pre-captured body templates. Extensive experiments on 4D-Dress, PuzzleIOI, and in-the-wild captures demonstrate that UP2You consistently surpasses previous methods in both geometric accuracy (Chamfer-15%, P2S-18% on PuzzleIOI) and texture fidelity (PSNR-21%, LPIPS-46% on 4D-Dress). UP2You is efficient (1.5 minutes per person), and versatile (supports arbitrary pose control, and training-free multi-garment 3D virtual try-on), making it practical for real-world scenarios where humans are casually captured. Both models and code will be released to facilitate future research on this underexplored task. Project Page: https://zcai0612.github.io/UP2You&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2509.24817</guid>
      <pubDate>Mon, 29 Sep 2025 14:06:00 +0000</pubDate>
    </item>
    <item>
      <title>OmniRetarget: Interaction-Preserving Data Generation for Humanoid   Whole-Body Loco-Manipulation and Scene Interaction</title>
      <link>https://arxiv.org/abs/2509.26633</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.26633.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Lujie Yang, Xiaoyu Huang, Zhen Wu, Angjoo Kanazawa, Pieter Abbeel, Carmelo Sferrazza, C. Karen Liu, Rocky Duan, Guanya Shi&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; A dominant paradigm for teaching humanoid robots complex skills is to retarget human motions as kinematic references to train reinforcement learning (RL) policies. However, existing retargeting pipelines often struggle with the significant embodiment gap between humans and robots, producing physically implausible artifacts like foot-skating and penetration. More importantly, common retargeting methods neglect the rich human-object and human-environment interactions essential for expressive locomotion and loco-manipulation. To address this, we introduce OmniRetarget, an interaction-preserving data generation engine based on an interaction mesh that explicitly models and preserves the crucial spatial and contact relationships between an agent, the terrain, and manipulated objects. By minimizing the Laplacian deformation between the human and robot meshes while enforcing kinematic constraints, OmniRetarget generates kinematically feasible trajectories. Moreover, preserving task-relevant interactions enables efficient data augmentation, from a single demonstration to different robot embodiments, terrains, and object configurations. We comprehensively evaluate OmniRetarget by retargeting motions from OMOMO, LAFAN1, and our in-house MoCap datasets, generating over 8-hour trajectories that achieve better kinematic constraint satisfaction and contact preservation than widely used baselines. Such high-quality data enables proprioceptive RL policies to successfully execute long-horizon (up to 30 seconds) parkour and loco-manipulation skills on a Unitree G1 humanoid, trained with only 5 reward terms and simple domain randomization shared by all tasks, without any learning curriculum.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2509.26633.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Lujie Yang, Xiaoyu Huang, Zhen Wu, Angjoo Kanazawa, Pieter Abbeel, Carmelo Sferrazza, C. Karen Liu, Rocky Duan, Guanya Shi&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 4&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; A dominant paradigm for teaching humanoid robots complex skills is to retarget human motions as kinematic references to train reinforcement learning (RL) policies. However, existing retargeting pipelines often struggle with the significant embodiment gap between humans and robots, producing physically implausible artifacts like foot-skating and penetration. More importantly, common retargeting methods neglect the rich human-object and human-environment interactions essential for expressive locomotion and loco-manipulation. To address this, we introduce OmniRetarget, an interaction-preserving data generation engine based on an interaction mesh that explicitly models and preserves the crucial spatial and contact relationships between an agent, the terrain, and manipulated objects. By minimizing the Laplacian deformation between the human and robot meshes while enforcing kinematic constraints, OmniRetarget generates kinematically feasible trajectories. Moreover, preserving task-relevant interactions enables efficient data augmentation, from a single demonstration to different robot embodiments, terrains, and object configurations. We comprehensively evaluate OmniRetarget by retargeting motions from OMOMO, LAFAN1, and our in-house MoCap datasets, generating over 8-hour trajectories that achieve better kinematic constraint satisfaction and contact preservation than widely used baselines. Such high-quality data enables proprioceptive RL policies to successfully execute long-horizon (up to 30 seconds) parkour and loco-manipulation skills on a Unitree G1 humanoid, trained with only 5 reward terms and simple domain randomization shared by all tasks, without any learning curriculum.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2509.26633</guid>
      <pubDate>Tue, 30 Sep 2025 17:59:02 +0000</pubDate>
    </item>
@@ -81,14 +81,14 @@
    <item>
      <title>Low-probability Tokens Sustain Exploration in Reinforcement Learning   with Verifiable Reward</title>
      <link>https://arxiv.org/abs/2510.03222</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03222.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Guanhua Huang, Tingqiang Xu, Mingze Wang, Qi Yi, Xue Gong, Siheng Li, Ruibin Xiong, Kejiao Li, Yuhao Jiang, Bo Zhou&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 14&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reinforcement Learning with Verifiable Rewards (RLVR) has propelled Large Language Models in complex reasoning, yet its scalability is often hindered by a training bottleneck where performance plateaus as policy entropy collapses, signaling a loss of exploration. Previous methods typically address this by maintaining high policy entropy, yet the precise mechanisms that govern meaningful exploration have remained underexplored. Our analysis suggests that an unselective focus on entropy risks amplifying irrelevant tokens and destabilizing training. This paper investigates the exploration dynamics within RLVR and identifies a key issue: the gradual elimination of valuable low-probability exploratory tokens, which we term \textit{reasoning sparks}. We find that while abundant in pre-trained models, these sparks are systematically extinguished during RLVR due to over-penalization, leading to a degeneracy in exploration. To address this, we introduce Low-probability Regularization (Lp-Reg). Its core mechanism regularizes the policy towards a heuristic proxy distribution. This proxy is constructed by filtering out presumed noise tokens and re-normalizing the distribution over the remaining candidates. The result is a less-noisy proxy where the probability of reasoning sparks is amplified, which then serves as a soft regularization target to shield these valuable tokens from elimination via KL divergence. Experiments show that Lp-Reg enables stable on-policy training for around 1,000 steps, a regime where baseline entropy-control methods collapse. This sustained exploration leads to state-of-the-art performance, achieving a 60.17% average accuracy on five math benchmarks, an improvement of 2.66% over prior methods. Code is available at https://github.com/CarlanLark/Lp-Reg.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03222.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Guanhua Huang, Tingqiang Xu, Mingze Wang, Qi Yi, Xue Gong, Siheng Li, Ruibin Xiong, Kejiao Li, Yuhao Jiang, Bo Zhou&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 34&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reinforcement Learning with Verifiable Rewards (RLVR) has propelled Large Language Models in complex reasoning, yet its scalability is often hindered by a training bottleneck where performance plateaus as policy entropy collapses, signaling a loss of exploration. Previous methods typically address this by maintaining high policy entropy, yet the precise mechanisms that govern meaningful exploration have remained underexplored. Our analysis suggests that an unselective focus on entropy risks amplifying irrelevant tokens and destabilizing training. This paper investigates the exploration dynamics within RLVR and identifies a key issue: the gradual elimination of valuable low-probability exploratory tokens, which we term \textit{reasoning sparks}. We find that while abundant in pre-trained models, these sparks are systematically extinguished during RLVR due to over-penalization, leading to a degeneracy in exploration. To address this, we introduce Low-probability Regularization (Lp-Reg). Its core mechanism regularizes the policy towards a heuristic proxy distribution. This proxy is constructed by filtering out presumed noise tokens and re-normalizing the distribution over the remaining candidates. The result is a less-noisy proxy where the probability of reasoning sparks is amplified, which then serves as a soft regularization target to shield these valuable tokens from elimination via KL divergence. Experiments show that Lp-Reg enables stable on-policy training for around 1,000 steps, a regime where baseline entropy-control methods collapse. This sustained exploration leads to state-of-the-art performance, achieving a 60.17% average accuracy on five math benchmarks, an improvement of 2.66% over prior methods. Code is available at https://github.com/CarlanLark/Lp-Reg.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.03222</guid>
      <pubDate>Fri, 03 Oct 2025 17:56:13 +0000</pubDate>
    </item>
    <item>
      <title>UNIDOC-BENCH: A Unified Benchmark for Document-Centric Multimodal RAG</title>
      <link>https://arxiv.org/abs/2510.03663</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03663.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiangyu Peng, Cab Qin, Zeyuan Chen, Ran Xu, Caiming Xiong, Chien-Sheng Wu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 13&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Multimodal retrieval-augmented generation (MM-RAG) is a key approach for applying large language models (LLMs) and agents to real-world knowledge bases, yet current evaluations are fragmented, focusing on either text or images in isolation or on simplified multimodal setups that fail to capture document-centric multimodal use cases. In this paper, we introduce UniDoc-Bench, the first large-scale, realistic benchmark for MM-RAG built from 70k real-world PDF pages across eight domains. Our pipeline extracts and links evidence from text, tables, and figures, then generates 1,600 multimodal QA pairs spanning factual retrieval, comparison, summarization, and logical reasoning queries. To ensure reliability, 20% of QA pairs are validated by multiple annotators and expert adjudication. UniDoc-Bench supports apples-to-apples comparison across four paradigms: (1) text-only, (2) image-only, (3) multimodal text-image fusion, and (4) multimodal joint retrieval -- under a unified protocol with standardized candidate pools, prompts, and evaluation metrics. Our experiments show that multimodal text-image fusion RAG systems consistently outperform both unimodal and jointly multimodal embedding-based retrieval, indicating that neither text nor images alone are sufficient and that current multimodal embeddings remain inadequate. Beyond benchmarking, our analysis reveals when and how visual context complements textual evidence, uncovers systematic failure modes, and offers actionable guidance for developing more robust MM-RAG pipelines.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.03663.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiangyu Peng, Cab Qin, Zeyuan Chen, Ran Xu, Caiming Xiong, Chien-Sheng Wu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 14&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Multimodal retrieval-augmented generation (MM-RAG) is a key approach for applying large language models (LLMs) and agents to real-world knowledge bases, yet current evaluations are fragmented, focusing on either text or images in isolation or on simplified multimodal setups that fail to capture document-centric multimodal use cases. In this paper, we introduce UniDoc-Bench, the first large-scale, realistic benchmark for MM-RAG built from 70k real-world PDF pages across eight domains. Our pipeline extracts and links evidence from text, tables, and figures, then generates 1,600 multimodal QA pairs spanning factual retrieval, comparison, summarization, and logical reasoning queries. To ensure reliability, 20% of QA pairs are validated by multiple annotators and expert adjudication. UniDoc-Bench supports apples-to-apples comparison across four paradigms: (1) text-only, (2) image-only, (3) multimodal text-image fusion, and (4) multimodal joint retrieval -- under a unified protocol with standardized candidate pools, prompts, and evaluation metrics. Our experiments show that multimodal text-image fusion RAG systems consistently outperform both unimodal and jointly multimodal embedding-based retrieval, indicating that neither text nor images alone are sufficient and that current multimodal embeddings remain inadequate. Beyond benchmarking, our analysis reveals when and how visual context complements textual evidence, uncovers systematic failure modes, and offers actionable guidance for developing more robust MM-RAG pipelines.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.03663</guid>
      <pubDate>Sat, 04 Oct 2025 04:30:13 +0000</pubDate>
    </item>
@@ -102,35 +102,35 @@
    <item>
      <title>DreamOmni2: Multimodal Instruction-based Editing and Generation</title>
      <link>https://arxiv.org/abs/2510.06679</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.06679.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Bin Xia, Bohao Peng, Yuechen Zhang, Junjia Huang, Jiyang Liu, Jingyao Li, Haoru Tan, Sitong Wu, Chengyao Wang, Yitong Wang, Xinglong Wu, Bei Yu, Jiaya Jia&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advancements in instruction-based image editing and subject-driven generation have garnered significant attention, yet both tasks still face limitations in meeting practical user needs. Instruction-based editing relies solely on language instructions, which often fail to capture specific editing details, making reference images necessary. Meanwhile, subject-driven generation is limited to combining concrete objects or people, overlooking broader, abstract concepts. To address these challenges, we propose two novel tasks: multimodal instruction-based editing and generation. These tasks support both text and image instructions and extend the scope to include both concrete and abstract concepts, greatly enhancing their practical applications. We introduce DreamOmni2, tackling two primary challenges: data creation and model framework design. Our data synthesis pipeline consists of three steps: (1) using a feature mixing method to create extraction data for both abstract and concrete concepts, (2) generating multimodal instruction-based editing training data using the editing and extraction models, and (3) further applying the extraction model to create training data for multimodal instruction-based editing. For the framework, to handle multi-image input, we propose an index encoding and position encoding shift scheme, which helps the model distinguish images and avoid pixel confusion. Additionally, we introduce joint training with the VLM and our generation/editing model to better process complex instructions. In addition, we have proposed comprehensive benchmarks for these two new tasks to drive their development. Experiments show that DreamOmni2 has achieved impressive results. Models and codes will be released.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.06679.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Bin Xia, Bohao Peng, Yuechen Zhang, Junjia Huang, Jiyang Liu, Jingyao Li, Haoru Tan, Sitong Wu, Chengyao Wang, Yitong Wang, Xinglong Wu, Bei Yu, Jiaya Jia&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 42&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advancements in instruction-based image editing and subject-driven generation have garnered significant attention, yet both tasks still face limitations in meeting practical user needs. Instruction-based editing relies solely on language instructions, which often fail to capture specific editing details, making reference images necessary. Meanwhile, subject-driven generation is limited to combining concrete objects or people, overlooking broader, abstract concepts. To address these challenges, we propose two novel tasks: multimodal instruction-based editing and generation. These tasks support both text and image instructions and extend the scope to include both concrete and abstract concepts, greatly enhancing their practical applications. We introduce DreamOmni2, tackling two primary challenges: data creation and model framework design. Our data synthesis pipeline consists of three steps: (1) using a feature mixing method to create extraction data for both abstract and concrete concepts, (2) generating multimodal instruction-based editing training data using the editing and extraction models, and (3) further applying the extraction model to create training data for multimodal instruction-based editing. For the framework, to handle multi-image input, we propose an index encoding and position encoding shift scheme, which helps the model distinguish images and avoid pixel confusion. Additionally, we introduce joint training with the VLM and our generation/editing model to better process complex instructions. In addition, we have proposed comprehensive benchmarks for these two new tasks to drive their development. Experiments show that DreamOmni2 has achieved impressive results. Models and codes will be released.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.06679</guid>
      <pubDate>Wed, 08 Oct 2025 06:07:14 +0000</pubDate>
    </item>
    <item>
      <title>LongRM: Revealing and Unlocking the Context Boundary of Reward Modeling</title>
      <link>https://arxiv.org/abs/2510.06915</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.06915.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Zecheng Tang, Baibei Ji, Quantong Qiu, Haitian Wang, Xiaobo Liang, Juntao Li, Min Zhang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 12&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reward model (RM) plays a pivotal role in aligning large language model (LLM) with human preferences. As real-world applications increasingly involve long history trajectories, e.g., LLM agent, it becomes indispensable to evaluate whether a model's responses are not only high-quality but also grounded in and consistent with the provided context. Yet, current RMs remain confined to short-context settings and primarily focus on response-level attributes (e.g., safety or helpfulness), while largely neglecting the critical dimension of long context-response consistency. In this work, we introduce Long-RewardBench, a benchmark specifically designed for long-context RM evaluation, featuring both Pairwise Comparison and Best-of-N tasks. Our preliminary study reveals that even state-of-the-art generative RMs exhibit significant fragility in long-context scenarios, failing to maintain context-aware preference judgments. Motivated by the analysis of failure patterns observed in model outputs, we propose a general multi-stage training strategy that effectively scales arbitrary models into robust Long-context RMs (LongRMs). Experiments show that our approach not only substantially improves performance on long-context evaluation but also preserves strong short-context capability. Notably, our 8B LongRM outperforms much larger 70B-scale baselines and matches the performance of the proprietary Gemini 2.5 Pro model.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.06915.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Zecheng Tang, Baibei Ji, Quantong Qiu, Haitian Wang, Xiaobo Liang, Juntao Li, Min Zhang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 13&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reward model (RM) plays a pivotal role in aligning large language model (LLM) with human preferences. As real-world applications increasingly involve long history trajectories, e.g., LLM agent, it becomes indispensable to evaluate whether a model's responses are not only high-quality but also grounded in and consistent with the provided context. Yet, current RMs remain confined to short-context settings and primarily focus on response-level attributes (e.g., safety or helpfulness), while largely neglecting the critical dimension of long context-response consistency. In this work, we introduce Long-RewardBench, a benchmark specifically designed for long-context RM evaluation, featuring both Pairwise Comparison and Best-of-N tasks. Our preliminary study reveals that even state-of-the-art generative RMs exhibit significant fragility in long-context scenarios, failing to maintain context-aware preference judgments. Motivated by the analysis of failure patterns observed in model outputs, we propose a general multi-stage training strategy that effectively scales arbitrary models into robust Long-context RMs (LongRMs). Experiments show that our approach not only substantially improves performance on long-context evaluation but also preserves strong short-context capability. Notably, our 8B LongRM outperforms much larger 70B-scale baselines and matches the performance of the proprietary Gemini 2.5 Pro model.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.06915</guid>
      <pubDate>Wed, 08 Oct 2025 11:48:16 +0000</pubDate>
    </item>
    <item>
      <title>Search-R3: Unifying Reasoning and Embedding Generation in Large Language   Models</title>
      <link>https://arxiv.org/abs/2510.07048</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07048.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yuntao Gui, James Cheng&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 0&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Despite their remarkable natural language understanding capabilities, Large Language Models (LLMs) have been underutilized for retrieval tasks. We present Search-R3, a novel framework that addresses this limitation by adapting LLMs to generate search embeddings as a direct output of their reasoning process. Our approach exploits LLMs' chain-of-thought capabilities, allowing them to produce more effective embeddings by reasoning step-by-step through complex semantic analyses. We implement this through three complementary mechanisms. (1) a supervised learning stage enables the model's ability to produce quality embeddings, (2) a reinforcement learning (RL) methodology that optimizes embedding generation alongside reasoning, and (3) a specialized RL environment that efficiently handles evolving embedding representations without requiring complete corpus re-encoding at each training iteration. Our extensive evaluations on diverse benchmarks demonstrate that Search-R3 significantly outperforms prior methods by unifying the reasoning and embedding generation processes. This integrated post-training approach represents a substantial advancement in handling complex knowledge-intensive tasks that require both sophisticated reasoning and effective information retrieval. Project page: https://github.com/ytgui/Search-R3&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07048.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yuntao Gui, James Cheng&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 3&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Despite their remarkable natural language understanding capabilities, Large Language Models (LLMs) have been underutilized for retrieval tasks. We present Search-R3, a novel framework that addresses this limitation by adapting LLMs to generate search embeddings as a direct output of their reasoning process. Our approach exploits LLMs' chain-of-thought capabilities, allowing them to produce more effective embeddings by reasoning step-by-step through complex semantic analyses. We implement this through three complementary mechanisms. (1) a supervised learning stage enables the model's ability to produce quality embeddings, (2) a reinforcement learning (RL) methodology that optimizes embedding generation alongside reasoning, and (3) a specialized RL environment that efficiently handles evolving embedding representations without requiring complete corpus re-encoding at each training iteration. Our extensive evaluations on diverse benchmarks demonstrate that Search-R3 significantly outperforms prior methods by unifying the reasoning and embedding generation processes. This integrated post-training approach represents a substantial advancement in handling complex knowledge-intensive tasks that require both sophisticated reasoning and effective information retrieval. Project page: https://github.com/ytgui/Search-R3&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07048</guid>
      <pubDate>Wed, 08 Oct 2025 14:16:20 +0000</pubDate>
    </item>
    <item>
      <title>NewtonBench: Benchmarking Generalizable Scientific Law Discovery in LLM   Agents</title>
      <link>https://arxiv.org/abs/2510.07172</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07172.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tianshi Zheng, Kelvin Kiu-Wai Tam, Newt Hue-Nam K. Nguyen, Baixuan Xu, Zhaowei Wang, Jiayang Cheng, Hong Ting Tsang, Weiqi Wang, Jiaxin Bai, Tianqing Fang, Yangqiu Song, Ginny Y. Wong, Simon See&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 25&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Large language models are emerging as powerful tools for scientific law discovery, a foundational challenge in AI-driven science. However, existing benchmarks for this task suffer from a fundamental methodological trilemma, forcing a trade-off between scientific relevance, scalability, and resistance to memorization. Furthermore, they oversimplify discovery as static function fitting, failing to capture the authentic scientific process of uncovering embedded laws through the interactive exploration of complex model systems. To address these critical gaps, we introduce NewtonBench, a benchmark comprising 324 scientific law discovery tasks across 12 physics domains. Our design mitigates the evaluation trilemma by using metaphysical shifts - systematic alterations of canonical laws - to generate a vast suite of problems that are scalable, scientifically relevant, and memorization-resistant. Moreover, we elevate the evaluation from static function fitting to interactive model discovery, requiring agents to experimentally probe simulated complex systems to uncover hidden principles. Our extensive experiment reveals a clear but fragile capability for discovery in frontier LLMs: this ability degrades precipitously with increasing system complexity and exhibits extreme sensitivity to observational noise. Notably, we uncover a paradoxical effect of tool assistance: providing a code interpreter can hinder more capable models by inducing a premature shift from exploration to exploitation, causing them to satisfice on suboptimal solutions. These results demonstrate that robust, generalizable discovery in complex, interactive environments remains the core challenge. By providing a scalable, robust, and scientifically authentic testbed, NewtonBench offers a crucial tool for measuring true progress and guiding the development of next-generation AI agents capable of genuine scientific discovery.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07172.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tianshi Zheng, Kelvin Kiu-Wai Tam, Newt Hue-Nam K. Nguyen, Baixuan Xu, Zhaowei Wang, Jiayang Cheng, Hong Ting Tsang, Weiqi Wang, Jiaxin Bai, Tianqing Fang, Yangqiu Song, Ginny Y. Wong, Simon See&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 27&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Large language models are emerging as powerful tools for scientific law discovery, a foundational challenge in AI-driven science. However, existing benchmarks for this task suffer from a fundamental methodological trilemma, forcing a trade-off between scientific relevance, scalability, and resistance to memorization. Furthermore, they oversimplify discovery as static function fitting, failing to capture the authentic scientific process of uncovering embedded laws through the interactive exploration of complex model systems. To address these critical gaps, we introduce NewtonBench, a benchmark comprising 324 scientific law discovery tasks across 12 physics domains. Our design mitigates the evaluation trilemma by using metaphysical shifts - systematic alterations of canonical laws - to generate a vast suite of problems that are scalable, scientifically relevant, and memorization-resistant. Moreover, we elevate the evaluation from static function fitting to interactive model discovery, requiring agents to experimentally probe simulated complex systems to uncover hidden principles. Our extensive experiment reveals a clear but fragile capability for discovery in frontier LLMs: this ability degrades precipitously with increasing system complexity and exhibits extreme sensitivity to observational noise. Notably, we uncover a paradoxical effect of tool assistance: providing a code interpreter can hinder more capable models by inducing a premature shift from exploration to exploitation, causing them to satisfice on suboptimal solutions. These results demonstrate that robust, generalizable discovery in complex, interactive environments remains the core challenge. By providing a scalable, robust, and scientifically authentic testbed, NewtonBench offers a crucial tool for measuring true progress and guiding the development of next-generation AI agents capable of genuine scientific discovery.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07172</guid>
      <pubDate>Wed, 08 Oct 2025 16:12:11 +0000</pubDate>
    </item>
    <item>
      <title>Hybrid Reinforcement: When Reward Is Sparse, It's Better to Be Dense</title>
      <link>https://arxiv.org/abs/2510.07242</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07242.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Leitian Tao, Ilia Kulikov, Swarnadeep Saha, Tianlu Wang, Jing Xu, Yixuan Li, Jason E Weston, Ping Yu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 26&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Post-training for reasoning of large language models (LLMs) increasingly relies on verifiable rewards: deterministic checkers that provide 0-1 correctness signals. While reliable, such binary feedback is brittle--many tasks admit partially correct or alternative answers that verifiers under-credit, and the resulting all-or-nothing supervision limits learning. Reward models offer richer, continuous feedback, which can serve as a complementary supervisory signal to verifiers. We introduce HERO (Hybrid Ensemble Reward Optimization), a reinforcement learning framework that integrates verifier signals with reward-model scores in a structured way. HERO employs stratified normalization to bound reward-model scores within verifier-defined groups, preserving correctness while refining quality distinctions, and variance-aware weighting to emphasize challenging prompts where dense signals matter most. Across diverse mathematical reasoning benchmarks, HERO consistently outperforms RM-only and verifier-only baselines, with strong gains on both verifiable and hard-to-verify tasks. Our results show that hybrid reward design retains the stability of verifiers while leveraging the nuance of reward models to advance reasoning.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07242.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Leitian Tao, Ilia Kulikov, Swarnadeep Saha, Tianlu Wang, Jing Xu, Yixuan Li, Jason E Weston, Ping Yu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 27&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Post-training for reasoning of large language models (LLMs) increasingly relies on verifiable rewards: deterministic checkers that provide 0-1 correctness signals. While reliable, such binary feedback is brittle--many tasks admit partially correct or alternative answers that verifiers under-credit, and the resulting all-or-nothing supervision limits learning. Reward models offer richer, continuous feedback, which can serve as a complementary supervisory signal to verifiers. We introduce HERO (Hybrid Ensemble Reward Optimization), a reinforcement learning framework that integrates verifier signals with reward-model scores in a structured way. HERO employs stratified normalization to bound reward-model scores within verifier-defined groups, preserving correctness while refining quality distinctions, and variance-aware weighting to emphasize challenging prompts where dense signals matter most. Across diverse mathematical reasoning benchmarks, HERO consistently outperforms RM-only and verifier-only baselines, with strong gains on both verifiable and hard-to-verify tasks. Our results show that hybrid reward design retains the stability of verifiers while leveraging the nuance of reward models to advance reasoning.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07242</guid>
      <pubDate>Wed, 08 Oct 2025 17:09:41 +0000</pubDate>
    </item>
@@ -151,28 +151,28 @@
    <item>
      <title>When Thoughts Meet Facts: Reusable Reasoning for Long-Context LMs</title>
      <link>https://arxiv.org/abs/2510.07499</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07499.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Soyeong Jeong, Taehee Jung, Sung Ju Hwang, Joo-Kyung Kim, Dongyeop Kang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 37&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent Long-Context Language Models (LCLMs) can process hundreds of thousands of tokens in a single prompt, enabling new opportunities for knowledge-intensive multi-hop reasoning by integrating large sets of retrieved documents or, in some cases, directly all necessary information. However, simply feeding more documents into the context window fails to capture how evidence should be connected. We address this gap with thought templates, which recast reasoning as reusable thought caches, derived from prior problem solving traces, structuring how evidence is combined and guiding multi-hop inference with factual documents. To keep these templates effective, we propose an update strategy that iteratively refines templates derived from training data through natural-language feedback. Across diverse benchmarks and LCLM families, our approach delivers consistent gains over strong baselines in both retrieval-based and retrieval-free settings. Furthermore, we show that optimized templates can be distilled into smaller open-source models, demonstrating its broad applicability and transparent reasoning reuse. We refer to our framework as Thought Template Augmented LCLMs (ToTAL).&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07499.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Soyeong Jeong, Taehee Jung, Sung Ju Hwang, Joo-Kyung Kim, Dongyeop Kang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 40&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent Long-Context Language Models (LCLMs) can process hundreds of thousands of tokens in a single prompt, enabling new opportunities for knowledge-intensive multi-hop reasoning by integrating large sets of retrieved documents or, in some cases, directly all necessary information. However, simply feeding more documents into the context window fails to capture how evidence should be connected. We address this gap with thought templates, which recast reasoning as reusable thought caches, derived from prior problem solving traces, structuring how evidence is combined and guiding multi-hop inference with factual documents. To keep these templates effective, we propose an update strategy that iteratively refines templates derived from training data through natural-language feedback. Across diverse benchmarks and LCLM families, our approach delivers consistent gains over strong baselines in both retrieval-based and retrieval-free settings. Furthermore, we show that optimized templates can be distilled into smaller open-source models, demonstrating its broad applicability and transparent reasoning reuse. We refer to our framework as Thought Template Augmented LCLMs (ToTAL).&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07499</guid>
      <pubDate>Wed, 08 Oct 2025 19:52:35 +0000</pubDate>
    </item>
    <item>
      <title>PickStyle: Video-to-Video Style Transfer with Context-Style Adapters</title>
      <link>https://arxiv.org/abs/2510.07546</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07546.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Soroush Mehraban, Vida Adeli, Jacob Rommann, Babak Taati, Kyryl Truskovskyi&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 15&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We address the task of video style transfer with diffusion models, where the goal is to preserve the context of an input video while rendering it in a target style specified by a text prompt. A major challenge is the lack of paired video data for supervision. We propose PickStyle, a video-to-video style transfer framework that augments pretrained video diffusion backbones with style adapters and benefits from paired still image data with source-style correspondences for training. PickStyle inserts low-rank adapters into the self-attention layers of conditioning modules, enabling efficient specialization for motion-style transfer while maintaining strong alignment between video content and style. To bridge the gap between static image supervision and dynamic video, we construct synthetic training clips from paired images by applying shared augmentations that simulate camera motion, ensuring temporal priors are preserved. In addition, we introduce Context-Style Classifier-Free Guidance (CS-CFG), a novel factorization of classifier-free guidance into independent text (style) and video (context) directions. CS-CFG ensures that context is preserved in generated video while the style is effectively transferred. Experiments across benchmarks show that our approach achieves temporally coherent, style-faithful, and content-preserving video translations, outperforming existing baselines both qualitatively and quantitatively.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07546.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Soroush Mehraban, Vida Adeli, Jacob Rommann, Babak Taati, Kyryl Truskovskyi&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 16&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We address the task of video style transfer with diffusion models, where the goal is to preserve the context of an input video while rendering it in a target style specified by a text prompt. A major challenge is the lack of paired video data for supervision. We propose PickStyle, a video-to-video style transfer framework that augments pretrained video diffusion backbones with style adapters and benefits from paired still image data with source-style correspondences for training. PickStyle inserts low-rank adapters into the self-attention layers of conditioning modules, enabling efficient specialization for motion-style transfer while maintaining strong alignment between video content and style. To bridge the gap between static image supervision and dynamic video, we construct synthetic training clips from paired images by applying shared augmentations that simulate camera motion, ensuring temporal priors are preserved. In addition, we introduce Context-Style Classifier-Free Guidance (CS-CFG), a novel factorization of classifier-free guidance into independent text (style) and video (context) directions. CS-CFG ensures that context is preserved in generated video while the style is effectively transferred. Experiments across benchmarks show that our approach achieves temporally coherent, style-faithful, and content-preserving video translations, outperforming existing baselines both qualitatively and quantitatively.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07546</guid>
      <pubDate>Wed, 08 Oct 2025 21:02:55 +0000</pubDate>
    </item>
    <item>
      <title>OpenRubrics: Towards Scalable Synthetic Rubric Generation for Reward   Modeling and LLM Alignment</title>
      <link>https://arxiv.org/abs/2510.07743</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07743.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tianci Liu, Ran Xu, Tony Yu, Ilgee Hong, Carl Yang, Tuo Zhao, Haoyu Wang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reward modeling lies at the core of reinforcement learning from human feedback (RLHF), yet most existing reward models rely on scalar or pairwise judgments that fail to capture the multifaceted nature of human preferences. Recent studies have explored rubrics-as-rewards (RaR) that uses structured natural language criteria that capture multiple dimensions of response quality. However, producing rubrics that are both reliable and scalable remains a key challenge. In this work, we introduce OpenRubrics, a diverse, large-scale collection of (prompt, rubric) pairs for training rubric-generation and rubric-based reward models. To elicit discriminative and comprehensive evaluation signals, we introduce Contrastive Rubric Generation (CRG), which derives both hard rules (explicit constraints) and principles (implicit qualities) by contrasting preferred and rejected responses. We further improve reliability by enforcing preference-label consistency via rejection sampling to remove noisy rubrics. Across multiple reward-modeling benchmarks, our rubric-based reward model, Rubric-RM, surpasses strong size-matched baselines by 6.8%. These gains transfer to policy models on instruction-following and biomedical benchmarks. Our results show that rubrics provide scalable alignment signals that narrow the gap between costly human evaluation and automated reward modeling, enabling a new principle-driven paradigm for LLM alignment.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07743.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tianci Liu, Ran Xu, Tony Yu, Ilgee Hong, Carl Yang, Tuo Zhao, Haoyu Wang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 7&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reward modeling lies at the core of reinforcement learning from human feedback (RLHF), yet most existing reward models rely on scalar or pairwise judgments that fail to capture the multifaceted nature of human preferences. Recent studies have explored rubrics-as-rewards (RaR) that uses structured natural language criteria that capture multiple dimensions of response quality. However, producing rubrics that are both reliable and scalable remains a key challenge. In this work, we introduce OpenRubrics, a diverse, large-scale collection of (prompt, rubric) pairs for training rubric-generation and rubric-based reward models. To elicit discriminative and comprehensive evaluation signals, we introduce Contrastive Rubric Generation (CRG), which derives both hard rules (explicit constraints) and principles (implicit qualities) by contrasting preferred and rejected responses. We further improve reliability by enforcing preference-label consistency via rejection sampling to remove noisy rubrics. Across multiple reward-modeling benchmarks, our rubric-based reward model, Rubric-RM, surpasses strong size-matched baselines by 6.8%. These gains transfer to policy models on instruction-following and biomedical benchmarks. Our results show that rubrics provide scalable alignment signals that narrow the gap between costly human evaluation and automated reward modeling, enabling a new principle-driven paradigm for LLM alignment.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07743</guid>
      <pubDate>Thu, 09 Oct 2025 03:31:26 +0000</pubDate>
    </item>
    <item>
      <title>GCPO: When Contrast Fails, Go Gold</title>
      <link>https://arxiv.org/abs/2510.07790</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07790.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Hao Wu, Wei Liu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 3&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reinforcement learning has been widely applied to enhance the reasoning capabilities of large language models. Extending the inference limits of smaller models has become a prominent research focus. However, algorithms such as Group Relative Policy Optimization (GRPO) suffer from a clear drawback: the upper bound of a model's rollout responses is entirely determined by the model itself, preventing the acquisition of knowledge from samples that are either all incorrect or all correct. In this paper, we introduce Group Contrastive Policy Optimization (GCPO), a method that incorporates external standard reference answers. When the model cannot solve a problem, the reference answer supplies the correct response, steering the model toward an unequivocally accurate update direction. This approach offers two main advantages: (1) it improves training efficiency by fully utilizing every sample; (2) it enables the model to emulate the problem solving strategy of the reference answer during training, thereby enhancing generalization in reasoning. GCPO achieves outstanding results across multiple benchmark datasets, yielding substantial improvements over the baseline model. Our code is available at: https://github.com/AchoWu/GCPO.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.07790.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Hao Wu, Wei Liu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 5&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Reinforcement learning has been widely applied to enhance the reasoning capabilities of large language models. Extending the inference limits of smaller models has become a prominent research focus. However, algorithms such as Group Relative Policy Optimization (GRPO) suffer from a clear drawback: the upper bound of a model's rollout responses is entirely determined by the model itself, preventing the acquisition of knowledge from samples that are either all incorrect or all correct. In this paper, we introduce Group Contrastive Policy Optimization (GCPO), a method that incorporates external standard reference answers. When the model cannot solve a problem, the reference answer supplies the correct response, steering the model toward an unequivocally accurate update direction. This approach offers two main advantages: (1) it improves training efficiency by fully utilizing every sample; (2) it enables the model to emulate the problem solving strategy of the reference answer during training, thereby enhancing generalization in reasoning. GCPO achieves outstanding results across multiple benchmark datasets, yielding substantial improvements over the baseline model. Our code is available at: https://github.com/AchoWu/GCPO.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.07790</guid>
      <pubDate>Thu, 09 Oct 2025 05:09:06 +0000</pubDate>
    </item>
@@ -186,7 +186,7 @@
    <item>
      <title>Learning on the Job: An Experience-Driven Self-Evolving Agent for   Long-Horizon Tasks</title>
      <link>https://arxiv.org/abs/2510.08002</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08002.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Cheng Yang, Xuemeng Yang, Licheng Wen, Daocheng Fu, Jianbiao Mei, Rong Wu, Pinlong Cai, Yufan Shen, Nianchen Deng, Botian Shi, Yu Qiao, Haifeng Li&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 9&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Large Language Models have demonstrated remarkable capabilities across diverse domains, yet significant challenges persist when deploying them as AI agents for real-world long-horizon tasks. Existing LLM agents suffer from a critical limitation: they are test-time static and cannot learn from experience, lacking the ability to accumulate knowledge and continuously improve on the job. To address this challenge, we propose MUSE, a novel agent framework that introduces an experience-driven, self-evolving system centered around a hierarchical Memory Module. MUSE organizes diverse levels of experience and leverages them to plan and execute long-horizon tasks across multiple applications. After each sub-task execution, the agent autonomously reflects on its trajectory, converting the raw trajectory into structured experience and integrating it back into the Memory Module. This mechanism enables the agent to evolve beyond its static pretrained parameters, fostering continuous learning and self-evolution. We evaluate MUSE on the long-horizon productivity benchmark TAC. It achieves new SOTA performance by a significant margin using only a lightweight Gemini-2.5 Flash model. Sufficient Experiments demonstrate that as the agent autonomously accumulates experience, it exhibits increasingly superior task completion capabilities, as well as robust continuous learning and self-evolution capabilities. Moreover, the accumulated experience from MUSE exhibits strong generalization properties, enabling zero-shot improvement on new tasks. MUSE establishes a new paradigm for AI agents capable of real-world productivity task automation.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08002.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Cheng Yang, Xuemeng Yang, Licheng Wen, Daocheng Fu, Jianbiao Mei, Rong Wu, Pinlong Cai, Yufan Shen, Nianchen Deng, Botian Shi, Yu Qiao, Haifeng Li&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 11&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Large Language Models have demonstrated remarkable capabilities across diverse domains, yet significant challenges persist when deploying them as AI agents for real-world long-horizon tasks. Existing LLM agents suffer from a critical limitation: they are test-time static and cannot learn from experience, lacking the ability to accumulate knowledge and continuously improve on the job. To address this challenge, we propose MUSE, a novel agent framework that introduces an experience-driven, self-evolving system centered around a hierarchical Memory Module. MUSE organizes diverse levels of experience and leverages them to plan and execute long-horizon tasks across multiple applications. After each sub-task execution, the agent autonomously reflects on its trajectory, converting the raw trajectory into structured experience and integrating it back into the Memory Module. This mechanism enables the agent to evolve beyond its static pretrained parameters, fostering continuous learning and self-evolution. We evaluate MUSE on the long-horizon productivity benchmark TAC. It achieves new SOTA performance by a significant margin using only a lightweight Gemini-2.5 Flash model. Sufficient Experiments demonstrate that as the agent autonomously accumulates experience, it exhibits increasingly superior task completion capabilities, as well as robust continuous learning and self-evolution capabilities. Moreover, the accumulated experience from MUSE exhibits strong generalization properties, enabling zero-shot improvement on new tasks. MUSE establishes a new paradigm for AI agents capable of real-world productivity task automation.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08002</guid>
      <pubDate>Thu, 09 Oct 2025 09:40:34 +0000</pubDate>
    </item>
@@ -207,7 +207,7 @@
    <item>
      <title>Training-Free Group Relative Policy Optimization</title>
      <link>https://arxiv.org/abs/2510.08191</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08191.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yuzheng Cai, Siqi Cai, Yuchen Shi, Zihan Xu, Lichao Chen, Yulei Qin, Xiaoyu Tan, Gang Li, Zongyi Li, Haojia Lin, Yong Mao, Ke Li, Xing Sun&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advances in Large Language Model (LLM) agents have demonstrated their promising general capabilities. However, their performance in specialized real-world domains often degrades due to challenges in effectively integrating external tools and specific prompting strategies. While methods like agentic reinforcement learning have been proposed to address this, they typically rely on costly parameter updates, for example, through a process that uses Supervised Fine-Tuning (SFT) followed by a Reinforcement Learning (RL) phase with Group Relative Policy Optimization (GRPO) to alter the output distribution. However, we argue that LLMs can achieve a similar effect on the output distribution by learning experiential knowledge as a token prior, which is a far more lightweight approach that not only addresses practical data scarcity but also avoids the common issue of overfitting. To this end, we propose Training-Free Group Relative Policy Optimization (Training-Free GRPO), a cost-effective solution that enhances LLM agent performance without any parameter updates. Our method leverages the group relative semantic advantage instead of numerical ones within each group of rollouts, iteratively distilling high-quality experiential knowledge during multi-epoch learning on a minimal ground-truth data. Such knowledge serves as the learned token prior, which is seamlessly integrated during LLM API calls to guide model behavior. Experiments on mathematical reasoning and web searching tasks demonstrate that Training-Free GRPO, when applied to DeepSeek-V3.1-Terminus, significantly improves out-of-domain performance. With just a few dozen training samples, Training-Free GRPO outperforms fine-tuned small LLMs with marginal training data and cost.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08191.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yuzheng Cai, Siqi Cai, Yuchen Shi, Zihan Xu, Lichao Chen, Yulei Qin, Xiaoyu Tan, Gang Li, Zongyi Li, Haojia Lin, Yong Mao, Ke Li, Xing Sun&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 30&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advances in Large Language Model (LLM) agents have demonstrated their promising general capabilities. However, their performance in specialized real-world domains often degrades due to challenges in effectively integrating external tools and specific prompting strategies. While methods like agentic reinforcement learning have been proposed to address this, they typically rely on costly parameter updates, for example, through a process that uses Supervised Fine-Tuning (SFT) followed by a Reinforcement Learning (RL) phase with Group Relative Policy Optimization (GRPO) to alter the output distribution. However, we argue that LLMs can achieve a similar effect on the output distribution by learning experiential knowledge as a token prior, which is a far more lightweight approach that not only addresses practical data scarcity but also avoids the common issue of overfitting. To this end, we propose Training-Free Group Relative Policy Optimization (Training-Free GRPO), a cost-effective solution that enhances LLM agent performance without any parameter updates. Our method leverages the group relative semantic advantage instead of numerical ones within each group of rollouts, iteratively distilling high-quality experiential knowledge during multi-epoch learning on a minimal ground-truth data. Such knowledge serves as the learned token prior, which is seamlessly integrated during LLM API calls to guide model behavior. Experiments on mathematical reasoning and web searching tasks demonstrate that Training-Free GRPO, when applied to DeepSeek-V3.1-Terminus, significantly improves out-of-domain performance. With just a few dozen training samples, Training-Free GRPO outperforms fine-tuned small LLMs with marginal training data and cost.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08191</guid>
      <pubDate>Thu, 09 Oct 2025 13:18:17 +0000</pubDate>
    </item>
@@ -221,7 +221,7 @@
    <item>
      <title>LLMs Learn to Deceive Unintentionally: Emergent Misalignment in   Dishonesty from Misaligned Samples to Biased Human-AI Interactions</title>
      <link>https://arxiv.org/abs/2510.08211</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08211.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; XuHao Hu, Peng Wang, Xiaoya Lu, Dongrui Liu, Xuanjing Huang, Jing Shao&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 18&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Previous research has shown that LLMs finetuned on malicious or incorrect completions within narrow domains (e.g., insecure code or incorrect medical advice) can become broadly misaligned to exhibit harmful behaviors, which is called emergent misalignment. In this work, we investigate whether this phenomenon can extend beyond safety behaviors to a broader spectrum of dishonesty and deception under high-stakes scenarios (e.g., lying under pressure and deceptive behavior). To explore this, we finetune open-sourced LLMs on misaligned completions across diverse domains. Experimental results demonstrate that LLMs show broadly misaligned behavior in dishonesty. Additionally, we further explore this phenomenon in a downstream combined finetuning setting, and find that introducing as little as 1% of misalignment data into a standard downstream task is sufficient to decrease honest behavior over 20%. Furthermore, we consider a more practical human-AI interaction environment where we simulate both benign and biased users to interact with the assistant LLM. Notably, we find that the assistant can be misaligned unintentionally to exacerbate its dishonesty with only 10% biased user population. In summary, we extend the study of emergent misalignment to the domain of dishonesty and deception under high-stakes scenarios, and demonstrate that this risk arises not only through direct finetuning, but also in downstream mixture tasks and practical human-AI interactions.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08211.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; XuHao Hu, Peng Wang, Xiaoya Lu, Dongrui Liu, Xuanjing Huang, Jing Shao&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Previous research has shown that LLMs finetuned on malicious or incorrect completions within narrow domains (e.g., insecure code or incorrect medical advice) can become broadly misaligned to exhibit harmful behaviors, which is called emergent misalignment. In this work, we investigate whether this phenomenon can extend beyond safety behaviors to a broader spectrum of dishonesty and deception under high-stakes scenarios (e.g., lying under pressure and deceptive behavior). To explore this, we finetune open-sourced LLMs on misaligned completions across diverse domains. Experimental results demonstrate that LLMs show broadly misaligned behavior in dishonesty. Additionally, we further explore this phenomenon in a downstream combined finetuning setting, and find that introducing as little as 1% of misalignment data into a standard downstream task is sufficient to decrease honest behavior over 20%. Furthermore, we consider a more practical human-AI interaction environment where we simulate both benign and biased users to interact with the assistant LLM. Notably, we find that the assistant can be misaligned unintentionally to exacerbate its dishonesty with only 10% biased user population. In summary, we extend the study of emergent misalignment to the domain of dishonesty and deception under high-stakes scenarios, and demonstrate that this risk arises not only through direct finetuning, but also in downstream mixture tasks and practical human-AI interactions.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08211</guid>
      <pubDate>Thu, 09 Oct 2025 13:35:19 +0000</pubDate>
    </item>
@@ -249,14 +249,14 @@
    <item>
      <title>First Try Matters: Revisiting the Role of Reflection in Reasoning Models</title>
      <link>https://arxiv.org/abs/2510.08308</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08308.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Liwei Kang, Yue Deng, Yao Xiao, Zhanfeng Mo, Wee Sun Lee, Lidong Bing&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 18&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Large language models have recently demonstrated significant gains in reasoning ability, often attributed to their capacity to generate longer chains of thought and engage in reflective reasoning. However, the contribution of reflections to performance improvement remains unclear. In this paper, we systematically analyze the rollouts of eight reasoning models on five mathematical datasets. We focus on reflective behaviours where the model has already produced an answer but continues reflecting before finalizing its output. Our analysis reveals that reflections are predominantly confirmatory and rarely alter the model's initial answer, a pattern consistent across models and datasets. To understand the role of reflections in training, we construct supervised fine-tuning (SFT) datasets with varying amounts of reflection steps. We observe that training models on rollouts with more reflection steps primarily enhances first-answer correctness rather than the ability to correct initially wrong answers through reflections. This motivates us to propose a question-aware early-stopping method that enhances inference-time token efficiency by stopping the reasoning process once a few plausible candidate answers are generated, thereby reducing unnecessary reflection steps. Motivated by this, we further propose to dynamically truncate the reflections after a candidate answer has appeared during generation, which reduces reasoning tokens by 24.5% across five mathematical datasets, within a 2.9% drop in accuracy.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08308.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Liwei Kang, Yue Deng, Yao Xiao, Zhanfeng Mo, Wee Sun Lee, Lidong Bing&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 22&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Large language models have recently demonstrated significant gains in reasoning ability, often attributed to their capacity to generate longer chains of thought and engage in reflective reasoning. However, the contribution of reflections to performance improvement remains unclear. In this paper, we systematically analyze the rollouts of eight reasoning models on five mathematical datasets. We focus on reflective behaviours where the model has already produced an answer but continues reflecting before finalizing its output. Our analysis reveals that reflections are predominantly confirmatory and rarely alter the model's initial answer, a pattern consistent across models and datasets. To understand the role of reflections in training, we construct supervised fine-tuning (SFT) datasets with varying amounts of reflection steps. We observe that training models on rollouts with more reflection steps primarily enhances first-answer correctness rather than the ability to correct initially wrong answers through reflections. This motivates us to propose a question-aware early-stopping method that enhances inference-time token efficiency by stopping the reasoning process once a few plausible candidate answers are generated, thereby reducing unnecessary reflection steps. Motivated by this, we further propose to dynamically truncate the reflections after a candidate answer has appeared during generation, which reduces reasoning tokens by 24.5% across five mathematical datasets, within a 2.9% drop in accuracy.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08308</guid>
      <pubDate>Thu, 09 Oct 2025 14:57:10 +0000</pubDate>
    </item>
    <item>
      <title>UniVideo: Unified Understanding, Generation, and Editing for Videos</title>
      <link>https://arxiv.org/abs/2510.08377</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08377.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Cong Wei, Quande Liu, Zixuan Ye, Qiulin Wang, Xintao Wang, Pengfei Wan, Kun Gai, Wenhu Chen&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 47&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Unified multimodal models have shown promising results in multimodal content generation and editing but remain largely limited to the image domain. In this work, we present UniVideo, a versatile framework that extends unified modeling to the video domain. UniVideo adopts a dual-stream design, combining a Multimodal Large Language Model (MLLM) for instruction understanding with a Multimodal DiT (MMDiT) for video generation. This design enables accurate interpretation of complex multimodal instructions while preserving visual consistency. Built on this architecture, UniVideo unifies diverse video generation and editing tasks under a single multimodal instruction paradigm and is jointly trained across them. Extensive experiments demonstrate that UniVideo matches or surpasses state-of-the-art task-specific baselines in text/image-to-video generation, in-context video generation and in-context video editing. Notably, the unified design of UniVideo enables two forms of generalization. First, UniVideo supports task composition, such as combining editing with style transfer, by integrating multiple capabilities within a single instruction. Second, even without explicit training on free-form video editing, UniVideo transfers its editing capability from large-scale image editing data to this setting, handling unseen instructions such as green-screening characters or changing materials within a video. Beyond these core capabilities, UniVideo also supports visual-prompt-based video generation, where the MLLM interprets visual prompts and guides the MMDiT during synthesis. To foster future research, we will release our model and code.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08377.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Cong Wei, Quande Liu, Zixuan Ye, Qiulin Wang, Xintao Wang, Pengfei Wan, Kun Gai, Wenhu Chen&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 51&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Unified multimodal models have shown promising results in multimodal content generation and editing but remain largely limited to the image domain. In this work, we present UniVideo, a versatile framework that extends unified modeling to the video domain. UniVideo adopts a dual-stream design, combining a Multimodal Large Language Model (MLLM) for instruction understanding with a Multimodal DiT (MMDiT) for video generation. This design enables accurate interpretation of complex multimodal instructions while preserving visual consistency. Built on this architecture, UniVideo unifies diverse video generation and editing tasks under a single multimodal instruction paradigm and is jointly trained across them. Extensive experiments demonstrate that UniVideo matches or surpasses state-of-the-art task-specific baselines in text/image-to-video generation, in-context video generation and in-context video editing. Notably, the unified design of UniVideo enables two forms of generalization. First, UniVideo supports task composition, such as combining editing with style transfer, by integrating multiple capabilities within a single instruction. Second, even without explicit training on free-form video editing, UniVideo transfers its editing capability from large-scale image editing data to this setting, handling unseen instructions such as green-screening characters or changing materials within a video. Beyond these core capabilities, UniVideo also supports visual-prompt-based video generation, where the MLLM interprets visual prompts and guides the MMDiT during synthesis. To foster future research, we will release our model and code.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08377</guid>
      <pubDate>Thu, 09 Oct 2025 16:01:30 +0000</pubDate>
    </item>
@@ -277,7 +277,7 @@
    <item>
      <title>DeepPrune: Parallel Scaling without Inter-trace Redundancy</title>
      <link>https://arxiv.org/abs/2510.08483</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08483.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Shangqing Tu, Yaxuan Li, Yushi Bai, Lei Hou, Juanzi Li&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Parallel scaling has emerged as a powerful paradigm to enhance reasoning capabilities in large language models (LLMs) by generating multiple Chain-of-Thought (CoT) traces simultaneously. However, this approach introduces significant computational inefficiency due to inter-trace redundancy -- our analysis reveals that over 80% of parallel reasoning traces yield identical final answers, representing substantial wasted computation. To address this critical efficiency bottleneck, we propose DeepPrune, a novel framework that enables efficient parallel scaling through dynamic pruning. Our method features a specialized judge model trained with focal loss and oversampling techniques to accurately predict answer equivalence from partial reasoning traces which realizes 0.87 AUROC on equivalence prediction, combined with an online greedy clustering algorithm that dynamically prunes redundant paths while preserving answer diversity. Comprehensive evaluations across three challenging benchmarks (AIME 2024, AIME 2025, and GPQA) and multiple reasoning models demonstrate that DeepPrune achieves remarkable token reduction by over 80% compared to conventional consensus sampling on most cases, while maintaining competitive accuracy within 3 percentage points. Our work establishes a new standard for efficient parallel reasoning, making high-performance reasoning more efficient. Our code and data are here: https://deepprune.github.io/&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08483.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Shangqing Tu, Yaxuan Li, Yushi Bai, Lei Hou, Juanzi Li&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 22&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Parallel scaling has emerged as a powerful paradigm to enhance reasoning capabilities in large language models (LLMs) by generating multiple Chain-of-Thought (CoT) traces simultaneously. However, this approach introduces significant computational inefficiency due to inter-trace redundancy -- our analysis reveals that over 80% of parallel reasoning traces yield identical final answers, representing substantial wasted computation. To address this critical efficiency bottleneck, we propose DeepPrune, a novel framework that enables efficient parallel scaling through dynamic pruning. Our method features a specialized judge model trained with focal loss and oversampling techniques to accurately predict answer equivalence from partial reasoning traces which realizes 0.87 AUROC on equivalence prediction, combined with an online greedy clustering algorithm that dynamically prunes redundant paths while preserving answer diversity. Comprehensive evaluations across three challenging benchmarks (AIME 2024, AIME 2025, and GPQA) and multiple reasoning models demonstrate that DeepPrune achieves remarkable token reduction by over 80% compared to conventional consensus sampling on most cases, while maintaining competitive accuracy within 3 percentage points. Our work establishes a new standard for efficient parallel reasoning, making high-performance reasoning more efficient. Our code and data are here: https://deepprune.github.io/&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08483</guid>
      <pubDate>Thu, 09 Oct 2025 17:24:54 +0000</pubDate>
    </item>
@@ -291,42 +291,42 @@
    <item>
      <title>CoMAS: Co-Evolving Multi-Agent Systems via Interaction Rewards</title>
      <link>https://arxiv.org/abs/2510.08529</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08529.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiangyuan Xue, Yifan Zhou, Guibin Zhang, Zaibin Zhang, Yijiang Li, Chen Zhang, Zhenfei Yin, Philip Torr, Wanli Ouyang, Lei Bai&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 13&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Self-evolution is a central research topic in enabling large language model (LLM)-based agents to continually improve their capabilities after pretraining. Recent research has witnessed a transition from reinforcement learning (RL)-free to RL-based methods. Current RL-based methods either rely on dense external reward signals or extract intrinsic reward signals from LLMs themselves. However, these approaches diverge from the self-evolution mechanisms observed in human intelligence, where individuals learn and improve through mutual discussion and collaboration. In this work, we introduce Co-Evolving Multi-Agent Systems (CoMAS), a novel framework that enables agents to improve autonomously by learning from inter-agent interactions without external supervision. CoMAS generates intrinsic rewards from rich discussion dynamics, employs an LLM-as-a-judge mechanism to formulate these rewards, and optimizes each agent's policy through RL, thereby enabling decentralized and scalable co-evolution. Experimental results demonstrate that CoMAS consistently outperforms untrained agents and achieves state-of-the-art performance across most evaluation settings. Ablation studies confirm the necessity of interaction-based reward signals and reveal promising scalability as the number and diversity of agents increase. These findings establish CoMAS as a novel and effective paradigm for self-evolution in LLM-based agents.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08529.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiangyuan Xue, Yifan Zhou, Guibin Zhang, Zaibin Zhang, Yijiang Li, Chen Zhang, Zhenfei Yin, Philip Torr, Wanli Ouyang, Lei Bai&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 16&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Self-evolution is a central research topic in enabling large language model (LLM)-based agents to continually improve their capabilities after pretraining. Recent research has witnessed a transition from reinforcement learning (RL)-free to RL-based methods. Current RL-based methods either rely on dense external reward signals or extract intrinsic reward signals from LLMs themselves. However, these approaches diverge from the self-evolution mechanisms observed in human intelligence, where individuals learn and improve through mutual discussion and collaboration. In this work, we introduce Co-Evolving Multi-Agent Systems (CoMAS), a novel framework that enables agents to improve autonomously by learning from inter-agent interactions without external supervision. CoMAS generates intrinsic rewards from rich discussion dynamics, employs an LLM-as-a-judge mechanism to formulate these rewards, and optimizes each agent's policy through RL, thereby enabling decentralized and scalable co-evolution. Experimental results demonstrate that CoMAS consistently outperforms untrained agents and achieves state-of-the-art performance across most evaluation settings. Ablation studies confirm the necessity of interaction-based reward signals and reveal promising scalability as the number and diversity of agents increase. These findings establish CoMAS as a novel and effective paradigm for self-evolution in LLM-based agents.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08529</guid>
      <pubDate>Thu, 09 Oct 2025 17:50:26 +0000</pubDate>
    </item>
    <item>
      <title>MM-HELIX: Boosting Multimodal Long-Chain Reflective Reasoning with   Holistic Platform and Adaptive Hybrid Policy Optimization</title>
      <link>https://arxiv.org/abs/2510.08540</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08540.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiangyu Zhao, Junming Lin, Tianhao Liang, Yifan Zhou, Wenhao Chai, Yuzhe Gu, Weiyun Wang, Kai Chen, Gen Luo, Wenwei Zhang, Junchi Yan, Hua Yang, Haodong Duan, Xue Yang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 89&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; While current Multimodal Large Language Models (MLLMs) have demonstrated proficiency in reasoning tasks such as mathematics and logic, their capacity for long-chain reflective reasoning, a prerequisite for solving complex real-world problems, remains largely underexplored. In this work, we first conduct an extensive empirical investigation to evaluate this capability. Leveraging a carefully designed data synthesis engine, we construct MM-HELIX, a multimodal benchmark consisting 1,260 samples of 42 challenging synthetic tasks that require iterative thinking and backtracking. Empirical results on this benchmark reveal that existing MLLMs exhibit significant performance deficits in long-chain reflective reasoning. To address this limitation, we generate post-training data and further explore learning paradigms for exploiting such data. We first develop the Step-Elicited Response Generation pipeline to create MM-HELIX-100K, a large-scale dataset of 100k high-quality, reflective reasoning traces for instruction-tuning stage. Given that standard Reinforcement Learning fails on complex tasks due to sparse reward signals and catastrophic forgetting after Supervised Fine-Tuning, we propose Adaptive Hybrid Policy Optimization (AHPO), a novel training strategy that dynamically unifies offline supervision and online optimization into a single stage. This strategy enables the model to learn from expert data when rewards are sparse and conduct independent exploration once proficient. When applied to the Qwen2.5-VL-7B baseline, our method achieves a +18.6\% accuracy improvement on MM-HELIX benchmark and demonstrates strong generalization with a +5.7\% average performance gain on general mathematic and logic tasks. Our work demonstrate that reflective reasoning in MLLMs can be effectively learned and generalized, paving the way for developing more capable MLLMs.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08540.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiangyu Zhao, Junming Lin, Tianhao Liang, Yifan Zhou, Wenhao Chai, Yuzhe Gu, Weiyun Wang, Kai Chen, Gen Luo, Wenwei Zhang, Junchi Yan, Hua Yang, Haodong Duan, Xue Yang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 96&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; While current Multimodal Large Language Models (MLLMs) have demonstrated proficiency in reasoning tasks such as mathematics and logic, their capacity for long-chain reflective reasoning, a prerequisite for solving complex real-world problems, remains largely underexplored. In this work, we first conduct an extensive empirical investigation to evaluate this capability. Leveraging a carefully designed data synthesis engine, we construct MM-HELIX, a multimodal benchmark consisting 1,260 samples of 42 challenging synthetic tasks that require iterative thinking and backtracking. Empirical results on this benchmark reveal that existing MLLMs exhibit significant performance deficits in long-chain reflective reasoning. To address this limitation, we generate post-training data and further explore learning paradigms for exploiting such data. We first develop the Step-Elicited Response Generation pipeline to create MM-HELIX-100K, a large-scale dataset of 100k high-quality, reflective reasoning traces for instruction-tuning stage. Given that standard Reinforcement Learning fails on complex tasks due to sparse reward signals and catastrophic forgetting after Supervised Fine-Tuning, we propose Adaptive Hybrid Policy Optimization (AHPO), a novel training strategy that dynamically unifies offline supervision and online optimization into a single stage. This strategy enables the model to learn from expert data when rewards are sparse and conduct independent exploration once proficient. When applied to the Qwen2.5-VL-7B baseline, our method achieves a +18.6\% accuracy improvement on MM-HELIX benchmark and demonstrates strong generalization with a +5.7\% average performance gain on general mathematic and logic tasks. Our work demonstrate that reflective reasoning in MLLMs can be effectively learned and generalized, paving the way for developing more capable MLLMs.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08540</guid>
      <pubDate>Thu, 09 Oct 2025 17:53:58 +0000</pubDate>
    </item>
    <item>
      <title>R2RGEN: Real-to-Real 3D Data Generation for Spatially Generalized   Manipulation</title>
      <link>https://arxiv.org/abs/2510.08547</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08547.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiuwei Xu, Angyuan Ma, Hankun Li, Bingyao Yu, Zheng Zhu, Jie Zhou, Jiwen Lu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Towards the aim of generalized robotic manipulation, spatial generalization is the most fundamental capability that requires the policy to work robustly under different spatial distribution of objects, environment and agent itself. To achieve this, substantial human demonstrations need to be collected to cover different spatial configurations for training a generalized visuomotor policy via imitation learning. Prior works explore a promising direction that leverages data generation to acquire abundant spatially diverse data from minimal source demonstrations. However, most approaches face significant sim-to-real gap and are often limited to constrained settings, such as fixed-base scenarios and predefined camera viewpoints. In this paper, we propose a real-to-real 3D data generation framework (R2RGen) that directly augments the pointcloud observation-action pairs to generate real-world data. R2RGen is simulator- and rendering-free, thus being efficient and plug-and-play. Specifically, given a single source demonstration, we introduce an annotation mechanism for fine-grained parsing of scene and trajectory. A group-wise augmentation strategy is proposed to handle complex multi-object compositions and diverse task constraints. We further present camera-aware processing to align the distribution of generated data with real-world 3D sensor. Empirically, R2RGen substantially enhances data efficiency on extensive experiments and demonstrates strong potential for scaling and application on mobile manipulation.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08547.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xiuwei Xu, Angyuan Ma, Hankun Li, Bingyao Yu, Zheng Zhu, Jie Zhou, Jiwen Lu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Towards the aim of generalized robotic manipulation, spatial generalization is the most fundamental capability that requires the policy to work robustly under different spatial distribution of objects, environment and agent itself. To achieve this, substantial human demonstrations need to be collected to cover different spatial configurations for training a generalized visuomotor policy via imitation learning. Prior works explore a promising direction that leverages data generation to acquire abundant spatially diverse data from minimal source demonstrations. However, most approaches face significant sim-to-real gap and are often limited to constrained settings, such as fixed-base scenarios and predefined camera viewpoints. In this paper, we propose a real-to-real 3D data generation framework (R2RGen) that directly augments the pointcloud observation-action pairs to generate real-world data. R2RGen is simulator- and rendering-free, thus being efficient and plug-and-play. Specifically, given a single source demonstration, we introduce an annotation mechanism for fine-grained parsing of scene and trajectory. A group-wise augmentation strategy is proposed to handle complex multi-object compositions and diverse task constraints. We further present camera-aware processing to align the distribution of generated data with real-world 3D sensor. Empirically, R2RGen substantially enhances data efficiency on extensive experiments and demonstrates strong potential for scaling and application on mobile manipulation.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08547</guid>
      <pubDate>Thu, 09 Oct 2025 17:55:44 +0000</pubDate>
    </item>
    <item>
      <title>Entropy Regularizing Activation: Boosting Continuous Control, Large   Language Models, and Image Classification with Activation as Entropy   Constraints</title>
      <link>https://arxiv.org/abs/2510.08549</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08549.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Zilin Kang, Chonghua Liao, Tingqiang Xu, Huazhe Xu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 5&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We propose ERA, a new paradigm that constrains the sampling entropy above given thresholds by applying specially designed activations to the outputs of models. Our approach demonstrates broad effectiveness across different domains: 1) for large language models(LLMs), boosting the AIME 2025 score for Qwen2.5-Math-7B by 37.4%; 2) for continuous control reinforcement learning agents, improving performance by more than 30% over strong baselines such as SAC on the challenging HumanoidBench; 3) for image classification, enhancing ImageNet top-1 accuracy by 0.69% for ResNet-50. These gains are achieved with a computational overhead of less than 7%. Our work validates output activation as a powerful tool for entropy control, opening a new direction for designing simpler and more robust algorithms.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08549.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Zilin Kang, Chonghua Liao, Tingqiang Xu, Huazhe Xu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 6&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We propose ERA, a new paradigm that constrains the sampling entropy above given thresholds by applying specially designed activations to the outputs of models. Our approach demonstrates broad effectiveness across different domains: 1) for large language models(LLMs), boosting the AIME 2025 score for Qwen2.5-Math-7B by 37.4%; 2) for continuous control reinforcement learning agents, improving performance by more than 30% over strong baselines such as SAC on the challenging HumanoidBench; 3) for image classification, enhancing ImageNet top-1 accuracy by 0.69% for ResNet-50. These gains are achieved with a computational overhead of less than 7%. Our work validates output activation as a powerful tool for entropy control, opening a new direction for designing simpler and more robust algorithms.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08549</guid>
      <pubDate>Thu, 09 Oct 2025 17:56:17 +0000</pubDate>
    </item>
    <item>
      <title>ARTDECO: Towards Efficient and High-Fidelity On-the-Fly 3D   Reconstruction with Structured Scene Representation</title>
      <link>https://arxiv.org/abs/2510.08551</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08551.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Guanghao Li, Kerui Ren, Linning Xu, Zhewen Zheng, Changjian Jiang, Xin Gao, Bo Dai, Jian Pu, Mulin Yu, Jiangmiao Pang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; On-the-fly 3D reconstruction from monocular image sequences is a long-standing challenge in computer vision, critical for applications such as real-to-sim, AR/VR, and robotics. Existing methods face a major tradeoff: per-scene optimization yields high fidelity but is computationally expensive, whereas feed-forward foundation models enable real-time inference but struggle with accuracy and robustness. In this work, we propose ARTDECO, a unified framework that combines the efficiency of feed-forward models with the reliability of SLAM-based pipelines. ARTDECO uses 3D foundation models for pose estimation and point prediction, coupled with a Gaussian decoder that transforms multi-scale features into structured 3D Gaussians. To sustain both fidelity and efficiency at scale, we design a hierarchical Gaussian representation with a LoD-aware rendering strategy, which improves rendering fidelity while reducing redundancy. Experiments on eight diverse indoor and outdoor benchmarks show that ARTDECO delivers interactive performance comparable to SLAM, robustness similar to feed-forward systems, and reconstruction quality close to per-scene optimization, providing a practical path toward on-the-fly digitization of real-world environments with both accurate geometry and high visual fidelity. Explore more demos on our project page: https://city-super.github.io/artdeco/.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08551.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Guanghao Li, Kerui Ren, Linning Xu, Zhewen Zheng, Changjian Jiang, Xin Gao, Bo Dai, Jian Pu, Mulin Yu, Jiangmiao Pang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 25&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; On-the-fly 3D reconstruction from monocular image sequences is a long-standing challenge in computer vision, critical for applications such as real-to-sim, AR/VR, and robotics. Existing methods face a major tradeoff: per-scene optimization yields high fidelity but is computationally expensive, whereas feed-forward foundation models enable real-time inference but struggle with accuracy and robustness. In this work, we propose ARTDECO, a unified framework that combines the efficiency of feed-forward models with the reliability of SLAM-based pipelines. ARTDECO uses 3D foundation models for pose estimation and point prediction, coupled with a Gaussian decoder that transforms multi-scale features into structured 3D Gaussians. To sustain both fidelity and efficiency at scale, we design a hierarchical Gaussian representation with a LoD-aware rendering strategy, which improves rendering fidelity while reducing redundancy. Experiments on eight diverse indoor and outdoor benchmarks show that ARTDECO delivers interactive performance comparable to SLAM, robustness similar to feed-forward systems, and reconstruction quality close to per-scene optimization, providing a practical path toward on-the-fly digitization of real-world environments with both accurate geometry and high visual fidelity. Explore more demos on our project page: https://city-super.github.io/artdeco/.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08551</guid>
      <pubDate>Thu, 09 Oct 2025 17:57:38 +0000</pubDate>
    </item>
    <item>
      <title>VideoCanvas: Unified Video Completion from Arbitrary Spatiotemporal   Patches via In-Context Conditioning</title>
      <link>https://arxiv.org/abs/2510.08555</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08555.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Minghong Cai, Qiulin Wang, Zongli Ye, Wenze Liu, Quande Liu, Weicai Ye, Xintao Wang, Pengfei Wan, Kun Gai, Xiangyu Yue&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 36&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We introduce the task of arbitrary spatio-temporal video completion, where a video is generated from arbitrary, user-specified patches placed at any spatial location and timestamp, akin to painting on a video canvas. This flexible formulation naturally unifies many existing controllable video generation tasks--including first-frame image-to-video, inpainting, extension, and interpolation--under a single, cohesive paradigm. Realizing this vision, however, faces a fundamental obstacle in modern latent video diffusion models: the temporal ambiguity introduced by causal VAEs, where multiple pixel frames are compressed into a single latent representation, making precise frame-level conditioning structurally difficult. We address this challenge with VideoCanvas, a novel framework that adapts the In-Context Conditioning (ICC) paradigm to this fine-grained control task with zero new parameters. We propose a hybrid conditioning strategy that decouples spatial and temporal control: spatial placement is handled via zero-padding, while temporal alignment is achieved through Temporal RoPE Interpolation, which assigns each condition a continuous fractional position within the latent sequence. This resolves the VAE's temporal ambiguity and enables pixel-frame-aware control on a frozen backbone. To evaluate this new capability, we develop VideoCanvasBench, the first benchmark for arbitrary spatio-temporal video completion, covering both intra-scene fidelity and inter-scene creativity. Experiments demonstrate that VideoCanvas significantly outperforms existing conditioning paradigms, establishing a new state of the art in flexible and unified video generation.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08555.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Minghong Cai, Qiulin Wang, Zongli Ye, Wenze Liu, Quande Liu, Weicai Ye, Xintao Wang, Pengfei Wan, Kun Gai, Xiangyu Yue&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 52&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We introduce the task of arbitrary spatio-temporal video completion, where a video is generated from arbitrary, user-specified patches placed at any spatial location and timestamp, akin to painting on a video canvas. This flexible formulation naturally unifies many existing controllable video generation tasks--including first-frame image-to-video, inpainting, extension, and interpolation--under a single, cohesive paradigm. Realizing this vision, however, faces a fundamental obstacle in modern latent video diffusion models: the temporal ambiguity introduced by causal VAEs, where multiple pixel frames are compressed into a single latent representation, making precise frame-level conditioning structurally difficult. We address this challenge with VideoCanvas, a novel framework that adapts the In-Context Conditioning (ICC) paradigm to this fine-grained control task with zero new parameters. We propose a hybrid conditioning strategy that decouples spatial and temporal control: spatial placement is handled via zero-padding, while temporal alignment is achieved through Temporal RoPE Interpolation, which assigns each condition a continuous fractional position within the latent sequence. This resolves the VAE's temporal ambiguity and enables pixel-frame-aware control on a frozen backbone. To evaluate this new capability, we develop VideoCanvasBench, the first benchmark for arbitrary spatio-temporal video completion, covering both intra-scene fidelity and inter-scene creativity. Experiments demonstrate that VideoCanvas significantly outperforms existing conditioning paradigms, establishing a new state of the art in flexible and unified video generation.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08555</guid>
      <pubDate>Thu, 09 Oct 2025 17:58:59 +0000</pubDate>
    </item>
@@ -340,7 +340,7 @@
    <item>
      <title>Agent Learning via Early Experience</title>
      <link>https://arxiv.org/abs/2510.08558</link>
-      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08558.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Kai Zhang, Xiangchao Chen, Bo Liu, Tianci Xue, Zeyi Liao, Zhihan Liu, Xiyao Wang, Yuting Ning, Zhaorun Chen, Xiaohan Fu, Jian Xie, Yuxuan Sun, Boyu Gou, Qi Qi, Zihang Meng, Jianwei Yang, Ning Zhang, Xian Li, Ashish Shah, Dat Huynh, Hengduo Li, Zi Yang, Sara Cao, Lawrence Jang, Shuyan Zhou, Jiacheng Zhu, Huan Sun, Jason Weston, Yu Su, Yifan Wu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 121&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; A long-term goal of language agents is to learn and improve through their own experience, ultimately outperforming humans in complex, real-world tasks. However, training agents from experience data with reinforcement learning remains difficult in many environments, which either lack verifiable rewards (e.g., websites) or require inefficient long-horizon rollouts (e.g., multi-turn tool use). As a result, most current agents rely on supervised fine-tuning on expert data, which is challenging to scale and generalizes poorly. This limitation stems from the nature of expert demonstrations: they capture only a narrow range of scenarios and expose the agent to limited environment diversity. We address this limitation with a middle-ground paradigm we call early experience: interaction data generated by the agent's own actions, where the resulting future states serve as supervision without reward signals. Within this paradigm we study two strategies of using such data: (1) Implicit world modeling, which uses collected states to ground the policy in environment dynamics; and (2) Self-reflection, where the agent learns from its suboptimal actions to improve reasoning and decision-making. We evaluate across eight diverse environments and multiple model families. Our approaches consistently improve effectiveness and out-of-domain generalization, highlighting the value of early experience. Moreover, in environments with verifiable rewards, our results provide promising signals that early experience offers a strong foundation for subsequent reinforcement learning, positioning it as a practical bridge between imitation learning and fully experience-driven agents.&lt;/p&gt;</description>
+      <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2510.08558.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Kai Zhang, Xiangchao Chen, Bo Liu, Tianci Xue, Zeyi Liao, Zhihan Liu, Xiyao Wang, Yuting Ning, Zhaorun Chen, Xiaohan Fu, Jian Xie, Yuxuan Sun, Boyu Gou, Qi Qi, Zihang Meng, Jianwei Yang, Ning Zhang, Xian Li, Ashish Shah, Dat Huynh, Hengduo Li, Zi Yang, Sara Cao, Lawrence Jang, Shuyan Zhou, Jiacheng Zhu, Huan Sun, Jason Weston, Yu Su, Yifan Wu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 153&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; A long-term goal of language agents is to learn and improve through their own experience, ultimately outperforming humans in complex, real-world tasks. However, training agents from experience data with reinforcement learning remains difficult in many environments, which either lack verifiable rewards (e.g., websites) or require inefficient long-horizon rollouts (e.g., multi-turn tool use). As a result, most current agents rely on supervised fine-tuning on expert data, which is challenging to scale and generalizes poorly. This limitation stems from the nature of expert demonstrations: they capture only a narrow range of scenarios and expose the agent to limited environment diversity. We address this limitation with a middle-ground paradigm we call early experience: interaction data generated by the agent's own actions, where the resulting future states serve as supervision without reward signals. Within this paradigm we study two strategies of using such data: (1) Implicit world modeling, which uses collected states to ground the policy in environment dynamics; and (2) Self-reflection, where the agent learns from its suboptimal actions to improve reasoning and decision-making. We evaluate across eight diverse environments and multiple model families. Our approaches consistently improve effectiveness and out-of-domain generalization, highlighting the value of early experience. Moreover, in environments with verifiable rewards, our results provide promising signals that early experience offers a strong foundation for subsequent reinforcement learning, positioning it as a practical bridge between imitation learning and fully experience-driven agents.&lt;/p&gt;</description>
      <guid isPermaLink="false">https://arxiv.org/abs/2510.08558</guid>
      <pubDate>Thu, 09 Oct 2025 17:59:17 +0000</pubDate>
    </item>