bot: update RSS feed
This commit is contained in:
github-actions[bot]
437
feed.xml
437
feed.xml
@@ -7,321 +7,286 @@
|
||||
<docs>http://www.rssboard.org/rss-specification</docs>
|
||||
<generator>python-feedgen</generator>
|
||||
<language>en</language>
|
||||
<lastBuildDate>Mon, 02 Feb 2026 00:03:29 +0000</lastBuildDate>
|
||||
<lastBuildDate>Tue, 03 Feb 2026 00:04:57 +0000</lastBuildDate>
|
||||
<item>
|
||||
<title>PRISM: Learning Design Knowledge from Data for Stylistic Design Improvement</title>
|
||||
<link>https://arxiv.org/abs/2601.11747</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.11747.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Huaxiaoyue Wang, Sunav Choudhary, Franck Dernoncourt, Yu Shen, Stefano Petrangeli</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> Graphic design often involves exploring different stylistic directions, which can be time-consuming for non-experts. We address this problem of stylistically improving designs based on natural language instructions. While VLMs have shown initial success in graphic design, their pretrained knowledge on styles is often too general and misaligned with specific domain data. For example, VLMs may associate minimalism with abstract designs, whereas designers emphasize shape and color choices. Our key insight is to leverage design data -- a collection of real-world designs that implicitly capture designer's principles -- to learn design knowledge and guide stylistic improvement. We propose PRISM (PRior-Informed Stylistic Modification) that constructs and applies a design knowledge base through three stages: (1) clustering high-variance designs to capture diversity within a style, (2) summarizing each cluster into actionable design knowledge, and (3) retrieving relevant knowledge during inference to enable style-aware improvement. Experiments on the Crello dataset show that PRISM achieves the highest average rank of 1.49 (closer to 1 is better) over baselines in style alignment. User studies further validate these results, showing that PRISM is consistently preferred by designers.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.11747</guid>
|
||||
<pubDate>Fri, 16 Jan 2026 19:56:13 +0000</pubDate>
|
||||
<title>RM -RF: Reward Model for Run-Free Unit Test Evaluation</title>
|
||||
<link>https://arxiv.org/abs/2601.13097</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.13097.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Elena Bruches, Daniil Grebenkin, Mikhail Klementev, Vadim Alperovich, Roman Derunets, Dari Baturova, Georgy Mkrtchyan, Oleg Sedukhin, Ivan Bondarenko, Nikolay Bushkov, Stanislav Moiseev</p><p><b>Upvotes:</b> 8</p><p><b>Summary:</b> We present RM-RF, a lightweight reward model for run-free evaluation of automatically generated unit tests. Instead of repeatedly compiling and executing candidate tests, RM-RF predicts - from source and test code alone - three execution-derived signals: (1) whether the augmented test suite compiles and runs successfully, (2) whether the generated test cases increase code coverage, and (3) whether the generated test cases improve the mutation kill rate. To train and evaluate RM-RF we assemble a multilingual dataset (Java, Python, Go) of focal files, test files, and candidate test additions labeled by an execution-based pipeline, and we release an associated dataset and methodology for comparative evaluation. We tested multiple model families and tuning regimes (zero-shot, full fine-tuning, and PEFT via LoRA), achieving an average F1 of 0.69 across the three targets. Compared to conventional compile-and-run instruments, RM-RF provides substantially lower latency and infrastructure cost while delivering competitive predictive fidelity, enabling fast, scalable feedback for large-scale test generation and RL-based code optimization.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.13097</guid>
|
||||
<pubDate>Mon, 19 Jan 2026 14:37:50 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>LoL: Longer than Longer, Scaling Video Generation to Hour</title>
|
||||
<link>https://arxiv.org/abs/2601.16914</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.16914.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Justin Cui, Jie Wu, Ming Li, Tao Yang, Xiaojie Li, Rui Wang, Andrew Bai, Yuanhao Ban, Cho-Jui Hsieh</p><p><b>Upvotes:</b> 15</p><p><b>Summary:</b> Recent research in long-form video generation has shifted from bidirectional to autoregressive models, yet these methods commonly suffer from error accumulation and a loss of long-term coherence. While attention sink frames have been introduced to mitigate this performance decay, they often induce a critical failure mode we term sink-collapse: the generated content repeatedly reverts to the sink frame, resulting in abrupt scene resets and cyclic motion patterns. Our analysis reveals that sink-collapse originates from an inherent conflict between the periodic structure of Rotary Position Embedding (RoPE) and the multi-head attention mechanisms prevalent in current generative models. To address it, we propose a lightweight, training-free approach that effectively suppresses this behavior by introducing multi-head RoPE jitter that breaks inter-head attention homogenization and mitigates long-horizon collapse. Extensive experiments show that our method successfully alleviates sink-collapse while preserving generation quality. To the best of our knowledge, this work achieves the first demonstration of real-time, streaming, and infinite-length video generation with little quality decay. As an illustration of this robustness, we generate continuous videos up to 12 hours in length, which, to our knowledge, is among the longest publicly demonstrated results in streaming video generation.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.16914</guid>
|
||||
<pubDate>Fri, 23 Jan 2026 17:21:35 +0000</pubDate>
|
||||
<title>Memorization Dynamics in Knowledge Distillation for Language Models</title>
|
||||
<link>https://arxiv.org/abs/2601.15394</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.15394.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jaydeep Borkar, Karan Chadha, Niloofar Mireshghallah, Yuchen Zhang, Irina-Elena Veliche, Archi Mitra, David A. Smith, Zheng Xu, Diego Garcia-Olano</p><p><b>Upvotes:</b> 2</p><p><b>Summary:</b> Knowledge Distillation (KD) is increasingly adopted to transfer capabilities from large language models to smaller ones, offering significant improvements in efficiency and utility while often surpassing standard fine-tuning. Beyond performance, KD is also explored as a privacy-preserving mechanism to mitigate the risk of training data leakage. While training data memorization has been extensively studied in standard pre-training and fine-tuning settings, its dynamics in a knowledge distillation setup remain poorly understood. In this work, we study memorization across the KD pipeline using three large language model (LLM) families (Pythia, OLMo-2, Qwen-3) and three datasets (FineWeb, Wikitext, Nemotron-CC-v2). We find: (1) distilled models memorize significantly less training data than standard fine-tuning (reducing memorization by more than 50%); (2) some examples are inherently easier to memorize and account for a large fraction of memorization during distillation (over ~95%); (3) student memorization is predictable prior to distillation using features based on zlib entropy, KL divergence, and perplexity; and (4) while soft and hard distillation have similar overall memorization rates, hard distillation poses a greater risk: it inherits 2.7times more teacher-specific examples than soft distillation. Overall, we demonstrate that distillation can provide both improved generalization and reduced memorization risks compared to standard fine-tuning.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.15394</guid>
|
||||
<pubDate>Wed, 21 Jan 2026 19:04:40 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Segment Length Matters: A Study of Segment Lengths on Audio Fingerprinting Performance</title>
|
||||
<link>https://arxiv.org/abs/2601.17690</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.17690.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ziling Gong, Yunyan Ouyang, Iram Kamdar, Melody Ma, Hongjie Chen, Franck Dernoncourt, Ryan A. Rossi, Nesreen K. Ahmed</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> Audio fingerprinting provides an identifiable representation of acoustic signals, which can be later used for identification and retrieval systems. To obtain a discriminative representation, the input audio is usually segmented into shorter time intervals, allowing local acoustic features to be extracted and analyzed. Modern neural approaches typically operate on short, fixed-duration audio segments, yet the choice of segment duration is often made heuristically and rarely examined in depth. In this paper, we study how segment length affects audio fingerprinting performance. We extend an existing neural fingerprinting architecture to adopt various segment lengths and evaluate retrieval accuracy across different segment lengths and query durations. Our results show that short segment lengths (0.5-second) generally achieve better performance. Moreover, we evaluate LLM capacity in recommending the best segment length, which shows that GPT-5-mini consistently gives the best suggestions across five considerations among three studied LLMs. Our findings provide practical guidance for selecting segment duration in large-scale neural audio retrieval systems.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.17690</guid>
|
||||
<pubDate>Sun, 25 Jan 2026 04:32:32 +0000</pubDate>
|
||||
<title>Robust Tool Use via Fission-GRPO: Learning to Recover from Execution Errors</title>
|
||||
<link>https://arxiv.org/abs/2601.15625</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.15625.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhiwei Zhang, Fei Zhao, Rui Wang, Zezhong Wang, Bin Liang, Jiakang Wang, Yao Hu, Shaosheng Cao, Kam-Fai Wong</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Large language models (LLMs) can call tools effectively, yet they remain brittle in multi-turn execution: following a tool call error, smaller models often degenerate into repetitive invalid re-invocations, failing to interpret error feedback and self-correct. This brittleness hinders reliable real-world deployment, where the execution errors are inherently inevitable during tool interaction procedures. We identify a key limitation of current approaches: standard reinforcement learning (RL) treats errors as sparse negative rewards, providing no guidance on how to recover, while pre-collected synthetic error-correction datasets suffer from distribution mismatch with the model's on-policy error modes. To bridge this gap, we propose Fission-GRPO, a framework that converts execution errors into corrective supervision within the RL training loop. Our core mechanism fissions each failed trajectory into a new training instance by augmenting it with diagnostic feedback from a finetuned Error Simulator, then resampling recovery rollouts on-policy. This enables the model to learn from the precise errors it makes during exploration, rather than from static, pre-collected error cases. On the BFCL v4 Multi-Turn, Fission-GRPO improves the error recovery rate of Qwen3-8B by 5.7% absolute, crucially, yielding a 4% overall accuracy gain (42.75% to 46.75%) over GRPO and outperforming specialized tool-use agents.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.15625</guid>
|
||||
<pubDate>Thu, 22 Jan 2026 03:57:35 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>EEG Foundation Models: Progresses, Benchmarking, and Open Problems</title>
|
||||
<link>https://arxiv.org/abs/2601.17883</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.17883.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Dingkun Liu, Yuheng Chen, Zhu Chen, Zhenyao Cui, Yaozhi Wen, Jiayu An, Jingwei Luo, Dongrui Wu</p><p><b>Upvotes:</b> 17</p><p><b>Summary:</b> Electroencephalography (EEG) foundation models have recently emerged as a promising paradigm for brain-computer interfaces (BCIs), aiming to learn transferable neural representations from large-scale heterogeneous recordings. Despite rapid progresses, there lacks fair and comprehensive comparisons of existing EEG foundation models, due to inconsistent pre-training objectives, preprocessing choices, and downstream evaluation protocols. This paper fills this gap. We first review 50 representative models and organize their design choices into a unified taxonomic framework including data standardization, model architectures, and self-supervised pre-training strategies. We then evaluate 12 open-source foundation models and competitive specialist baselines across 13 EEG datasets spanning nine BCI paradigms. Emphasizing real-world deployments, we consider both cross-subject generalization under a leave-one-subject-out protocol and rapid calibration under a within-subject few-shot setting. We further compare full-parameter fine-tuning with linear probing to assess the transferability of pre-trained representations, and examine the relationship between model scale and downstream performance. Our results indicate that: 1) linear probing is frequently insufficient; 2) specialist models trained from scratch remain competitive across many tasks; and, 3) larger foundation models do not necessarily yield better generalization performance under current data regimes and training practices.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.17883</guid>
|
||||
<pubDate>Sun, 25 Jan 2026 15:28:50 +0000</pubDate>
|
||||
<title>TAM-Eval: Evaluating LLMs for Automated Unit Test Maintenance</title>
|
||||
<link>https://arxiv.org/abs/2601.18241</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.18241.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Elena Bruches, Vadim Alperovich, Dari Baturova, Roman Derunets, Daniil Grebenkin, Georgy Mkrtchyan, Oleg Sedukhin, Mikhail Klementev, Ivan Bondarenko, Nikolay Bushkov, Stanislav Moiseev</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> While Large Language Models (LLMs) have shown promise in software engineering, their application to unit testing remains largely confined to isolated test generation or oracle prediction, neglecting the broader challenge of test suite maintenance. We introduce TAM-Eval (Test Automated Maintenance Evaluation), a framework and benchmark designed to evaluate model performance across three core test maintenance scenarios: creation, repair, and updating of test suites. Unlike prior work limited to function-level tasks, TAM-Eval operates at the test file level, while maintaining access to full repository context during isolated evaluation, better reflecting real-world maintenance workflows. Our benchmark comprises 1,539 automatically extracted and validated scenarios from Python, Java, and Go projects. TAM-Eval supports system-agnostic evaluation of both raw LLMs and agentic workflows, using a reference-free protocol based on test suite pass rate, code coverage, and mutation testing. Empirical results indicate that state-of-the-art LLMs have limited capabilities in realistic test maintenance processes and yield only marginal improvements in test effectiveness. We release TAM-Eval as an open-source framework to support future research in automated software testing. Our data and code are publicly available at https://github.com/trndcenter/TAM-Eval.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.18241</guid>
|
||||
<pubDate>Mon, 26 Jan 2026 07:47:22 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Flow-based Extremal Mathematical Structure Discovery</title>
|
||||
<link>https://arxiv.org/abs/2601.18005</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.18005.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Gergely Bérczi, Baran Hashemi, Jonas Klüver</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> The discovery of extremal structures in mathematics requires navigating vast and nonconvex landscapes where analytical methods offer little guidance and brute-force search becomes intractable. We introduce FlowBoost, a closed-loop generative framework that learns to discover rare and extremal geometric structures by combining three components: (i) a geometry-aware conditional flow-matching model that learns to sample high-quality configurations, (ii) reward-guided policy optimization with action exploration that directly optimizes the generation process toward the objective while maintaining diversity, and (iii) stochastic local search for both training-data generation and final refinement. Unlike prior open-loop approaches, such as PatternBoost that retrains on filtered discrete samples, or AlphaEvolve which relies on frozen Large Language Models (LLMs) as evolutionary mutation operators, FlowBoost enforces geometric feasibility during sampling, and propagates reward signal directly into the generative model, closing the optimization loop and requiring much smaller training sets and shorter training times, and reducing the required outer-loop iterations by orders of magnitude, while eliminating dependence on LLMs. We demonstrate the framework on four geometric optimization problems: sphere packing in hypercubes, circle packing maximizing sum of radii, the Heilbronn triangle problem, and star discrepancy minimization. In several cases, FlowBoost discovers configurations that match or exceed the best known results. For circle packings, we improve the best known lower bounds, surpassing the LLM-based system AlphaEvolve while using substantially fewer computational resources.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.18005</guid>
|
||||
<pubDate>Sun, 25 Jan 2026 21:41:47 +0000</pubDate>
|
||||
<title>DenseGRPO: From Sparse to Dense Reward for Flow Matching Model Alignment</title>
|
||||
<link>https://arxiv.org/abs/2601.20218</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20218.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haoyou Deng, Keyu Yan, Chaojie Mao, Xiang Wang, Yu Liu, Changxin Gao, Nong Sang</p><p><b>Upvotes:</b> 10</p><p><b>Summary:</b> Recent GRPO-based approaches built on flow matching models have shown remarkable improvements in human preference alignment for text-to-image generation. Nevertheless, they still suffer from the sparse reward problem: the terminal reward of the entire denoising trajectory is applied to all intermediate steps, resulting in a mismatch between the global feedback signals and the exact fine-grained contributions at intermediate denoising steps. To address this issue, we introduce DenseGRPO, a novel framework that aligns human preference with dense rewards, which evaluates the fine-grained contribution of each denoising step. Specifically, our approach includes two key components: (1) we propose to predict the step-wise reward gain as dense reward of each denoising step, which applies a reward model on the intermediate clean images via an ODE-based approach. This manner ensures an alignment between feedback signals and the contributions of individual steps, facilitating effective training; and (2) based on the estimated dense rewards, a mismatch drawback between the uniform exploration setting and the time-varying noise intensity in existing GRPO-based methods is revealed, leading to an inappropriate exploration space. Thus, we propose a reward-aware scheme to calibrate the exploration space by adaptively adjusting a timestep-specific stochasticity injection in the SDE sampler, ensuring a suitable exploration space at all timesteps. Extensive experiments on multiple standard benchmarks demonstrate the effectiveness of the proposed DenseGRPO and highlight the critical role of the valid dense rewards in flow matching model alignment.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20218</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 03:39:05 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Typhoon-S: Minimal Open Post-Training for Sovereign Large Language Models</title>
|
||||
<link>https://arxiv.org/abs/2601.18129</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.18129.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Kunat Pipatanakul, Pittawat Taveekitworachai</p><p><b>Upvotes:</b> 9</p><p><b>Summary:</b> Large language models (LLMs) have progressed rapidly; however, most state-of-the-art models are trained and evaluated primarily in high-resource languages such as English and Chinese, and are often developed by a small number of organizations with access to large-scale compute and data. This gatekeeping creates a practical barrier for sovereign settings in which a regional- or national-scale institution or domain owner must retain control and understanding of model weights, training data, and deployment while operating under limited resources and strict transparency constraints. To this end, we identify two core requirements: (1) adoptability, the ability to transform a base model into a general-purpose assistant, and (2) sovereign capability, the ability to perform high-stakes, region-specific tasks (e.g., legal reasoning in local languages and cultural knowledge). We investigate whether these requirements can be achieved without scaling massive instruction corpora or relying on complex preference tuning pipelines and large-scale reinforcement fine-tuning (RFT). We present Typhoon S, a minimal and open post-training recipe that combines supervised fine-tuning, on-policy distillation, and small-scale RFT. Using Thai as a representative case study, we demonstrate that our approach transforms both sovereign-adapted and general-purpose base models into instruction-tuned models with strong general performance. We further show that small-scale RFT with InK-GRPO -- an extension of GRPO that augments the GRPO loss with a next-word prediction loss -- improves Thai legal reasoning and Thai-specific knowledge while preserving general capabilities. Our results suggest that a carefully designed post-training strategy can reduce the required scale of instruction data and computation, providing a practical path toward high-quality sovereign LLMs under academic-scale resources.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.18129</guid>
|
||||
<pubDate>Mon, 26 Jan 2026 04:20:59 +0000</pubDate>
|
||||
<title>Continual GUI Agents</title>
|
||||
<link>https://arxiv.org/abs/2601.20732</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20732.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ziwei Liu, Borui Kang, Hangjie Yuan, Zixiang Zhao, Wei Li, Yifan Zhu, Tao Feng</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> As digital environments (data distribution) are in flux, with new GUI data arriving over time-introducing new domains or resolutions-agents trained on static environments deteriorate in performance. In this work, we introduce Continual GUI Agents, a new task that requires GUI agents to perform continual learning under shifted domains and resolutions. We find existing methods fail to maintain stable grounding as GUI distributions shift over time, due to the diversity of UI interaction points and regions in fluxing scenarios. To address this, we introduce GUI-Anchoring in Flux (GUI-AiF), a new reinforcement fine-tuning framework that stabilizes continual learning through two novel rewards: Anchoring Point Reward in Flux (APR-iF) and Anchoring Region Reward in Flux (ARR-iF). These rewards guide the agents to align with shifting interaction points and regions, mitigating the tendency of existing reward strategies to over-adapt to static grounding cues (e.g., fixed coordinates or element scales). Extensive experiments show GUI-AiF surpasses state-of-the-art baselines. Our work establishes the first continual learning framework for GUI agents, revealing the untapped potential of reinforcement fine-tuning for continual GUI Agents.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20732</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 16:06:31 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>FROST: Filtering Reasoning Outliers with Attention for Efficient Reasoning</title>
|
||||
<link>https://arxiv.org/abs/2601.19001</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.19001.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haozheng Luo, Zhuolin Jiang, Md Zahid Hasan, Yan Chen, Soumalya Sarkar</p><p><b>Upvotes:</b> 2</p><p><b>Summary:</b> We propose FROST, an attention-aware method for efficient reasoning. Unlike traditional approaches, FROST leverages attention weights to prune uncritical reasoning paths, yielding shorter and more reliable reasoning trajectories. Methodologically, we introduce the concept of reasoning outliers and design an attention-based mechanism to remove them. Theoretically, FROST preserves and enhances the model's reasoning capacity while eliminating outliers at the sentence level. Empirically, we validate FROST on four benchmarks using two strong reasoning models (Phi-4-Reasoning and GPT-OSS-20B), outperforming state-of-the-art methods such as TALE and ThinkLess. Notably, FROST achieves an average 69.68% reduction in token usage and a 26.70% improvement in accuracy over the base model. Furthermore, in evaluations of attention outlier metrics, FROST reduces the maximum infinity norm by 15.97% and the average kurtosis by 91.09% compared to the base model. Code is available at https://github.com/robinzixuan/FROST</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.19001</guid>
|
||||
<pubDate>Mon, 26 Jan 2026 22:23:09 +0000</pubDate>
|
||||
<title>Do Reasoning Models Enhance Embedding Models?</title>
|
||||
<link>https://arxiv.org/abs/2601.21192</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21192.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Wun Yu Chan, Shaojin Chen, Huihao Jing, Kwun Hang Lau, Elton Chun-Chai Li, Zihao Wang, Haoran Li, Yangqiu Song</p><p><b>Upvotes:</b> 22</p><p><b>Summary:</b> State-of-the-art embedding models are increasingly derived from decoder-only Large Language Model (LLM) backbones adapted via contrastive learning. Given the emergence of reasoning models trained via Reinforcement Learning with Verifiable Rewards (RLVR), a natural question arises: do enhanced reasoning translate to superior semantic representations when these models serve as embedding initializations? Contrary to expectation, our evaluation on MTEB and BRIGHT reveals a **null effect**: embedding models initialized from RLVR-tuned backbones yield no consistent performance advantage over their base counterparts when subjected to identical training recipes. To unpack this paradox, we introduce **H**ierarchical **R**epresentation **S**imilarity **A**nalysis (HRSA), a framework that decomposes similarity across representation, geometry, and function levels. HRSA reveals that while RLVR induces irreversible latent manifold's local geometry reorganization and reversible coordinate basis drift, it preserves the global manifold geometry and linear readout. Consequently, subsequent contrastive learning drives strong alignment between base- and reasoning-initialized models, a phenomenon we term **Manifold Realignment**. Empirically, our findings suggest that unlike Supervised Fine-Tuning (SFT), RLVR optimizes trajectories within an existing semantic landscape rather than fundamentally restructuring the landscape itself.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21192</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 02:48:34 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Benchmarking Reward Hack Detection in Code Environments via Contrastive Analysis</title>
|
||||
<link>https://arxiv.org/abs/2601.20103</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20103.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Darshan Deshpande, Anand Kannappan, Rebecca Qian</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> Recent advances in reinforcement learning for code generation have made robust environments essential to prevent reward hacking. As LLMs increasingly serve as evaluators in code-based RL, their ability to detect reward hacking remains understudied. In this paper, we propose a novel taxonomy of reward exploits spanning across 54 categories and introduce TRACE (Testing Reward Anomalies in Code Environments), a synthetically curated and human-verified benchmark containing 517 testing trajectories. Unlike prior work that evaluates reward hack detection in isolated classification scenarios, we contrast these evaluations with a more realistic, contrastive anomaly detection setup on TRACE. Our experiments reveal that models capture reward hacks more effectively in contrastive settings than in isolated classification settings, with GPT-5.2 with highest reasoning mode achieving the best detection rate at 63%, up from 45% in isolated settings on TRACE. Building on this insight, we demonstrate that state-of-the-art models struggle significantly more with semantically contextualized reward hacks compared to syntactically contextualized ones. We further conduct qualitative analyses of model behaviors, as well as ablation studies showing that the ratio of benign to hacked trajectories and analysis cluster sizes substantially impact detection performance. We release the benchmark and evaluation harness to enable the community to expand TRACE and evaluate their models.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20103</guid>
|
||||
<pubDate>Tue, 27 Jan 2026 22:45:43 +0000</pubDate>
|
||||
<title>Latent Chain-of-Thought as Planning: Decoupling Reasoning from Verbalization</title>
|
||||
<link>https://arxiv.org/abs/2601.21358</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21358.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiecong Wang, Hao Peng, Chunyang Liu</p><p><b>Upvotes:</b> 5</p><p><b>Summary:</b> Chain-of-Thought (CoT) empowers Large Language Models (LLMs) to tackle complex problems, but remains constrained by the computational cost and reasoning path collapse when grounded in discrete token spaces. Recent latent reasoning approaches attempt to optimize efficiency by performing reasoning within continuous hidden states. However, these methods typically operate as opaque end-to-end mappings from explicit reasoning steps to latent states, and often require a pre-defined number of latent steps during inference. In this work, we introduce PLaT (Planning with Latent Thoughts), a framework that reformulates latent reasoning as planning by fundamentally decouple reasoning from verbalization. We model reasoning as a deterministic trajectory of latent planning states, while a separate Decoder grounds these thoughts into text when necessary. This decoupling allows the model to dynamically determine when to terminate reasoning rather than relying on fixed hyperparameters. Empirical results on mathematical benchmarks reveal a distinct trade-off: while PLaT achieves lower greedy accuracy than baselines, it demonstrates superior scalability in terms of reasoning diversity. This indicates that PLaT learns a robust, broader solution space, offering a transparent and scalable foundation for inference-time search.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21358</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 07:38:18 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models</title>
|
||||
<link>https://arxiv.org/abs/2601.20354</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20354.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zengbin Wang, Xuecai Hu, Yong Wang, Feng Xiong, Man Zhang, Xiangxiang Chu</p><p><b>Upvotes:</b> 104</p><p><b>Summary:</b> Text-to-image (T2I) models have achieved remarkable success in generating high-fidelity images, but they often fail in handling complex spatial relationships, e.g., spatial perception, reasoning, or interaction. These critical aspects are largely overlooked by current benchmarks due to their short or information-sparse prompt design. In this paper, we introduce SpatialGenEval, a new benchmark designed to systematically evaluate the spatial intelligence of T2I models, covering two key aspects: (1) SpatialGenEval involves 1,230 long, information-dense prompts across 25 real-world scenes. Each prompt integrates 10 spatial sub-domains and corresponding 10 multi-choice question-answer pairs, ranging from object position and layout to occlusion and causality. Our extensive evaluation of 21 state-of-the-art models reveals that higher-order spatial reasoning remains a primary bottleneck. (2) To demonstrate that the utility of our information-dense design goes beyond simple evaluation, we also construct the SpatialT2I dataset. It contains 15,400 text-image pairs with rewritten prompts to ensure image consistency while preserving information density. Fine-tuned results on current foundation models (i.e., Stable Diffusion-XL, Uniworld-V1, OmniGen2) yield consistent performance gains (+4.2%, +5.7%, +4.4%) and more realistic effects in spatial relations, highlighting a data-centric paradigm to achieve spatial intelligence in T2I models.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20354</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 08:15:00 +0000</pubDate>
|
||||
<title>Revisiting Diffusion Model Predictions Through Dimensionality</title>
|
||||
<link>https://arxiv.org/abs/2601.21419</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21419.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Qing Jin, Chaoyang Wang</p><p><b>Upvotes:</b> 4</p><p><b>Summary:</b> Recent advances in diffusion and flow matching models have highlighted a shift in the preferred prediction target -- moving from noise (varepsilon) and velocity (v) to direct data (x) prediction -- particularly in high-dimensional settings. However, a formal explanation of why the optimal target depends on the specific properties of the data remains elusive. In this work, we provide a theoretical framework based on a generalized prediction formulation that accommodates arbitrary output targets, of which varepsilon-, v-, and x-prediction are special cases. We derive the analytical relationship between data's geometry and the optimal prediction target, offering a rigorous justification for why x-prediction becomes superior when the ambient dimension significantly exceeds the data's intrinsic dimension. Furthermore, while our theory identifies dimensionality as the governing factor for the optimal prediction target, the intrinsic dimension of manifold-bound data is typically intractable to estimate in practice. To bridge this gap, we propose k-Diff, a framework that employs a data-driven approach to learn the optimal prediction parameter k directly from data, bypassing the need for explicit dimension estimation. Extensive experiments in both latent-space and pixel-space image generation demonstrate that k-Diff consistently outperforms fixed-target baselines across varying architectures and data scales, providing a principled and automated approach to enhancing generative performance.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21419</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 08:56:55 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>STORM: Slot-based Task-aware Object-centric Representation for robotic Manipulation</title>
|
||||
<link>https://arxiv.org/abs/2601.20381</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20381.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Alexandre Chapin, Emmanuel Dellandréa, Liming Chen</p><p><b>Upvotes:</b> 0</p><p><b>Summary:</b> Visual foundation models provide strong perceptual features for robotics, but their dense representations lack explicit object-level structure, limiting robustness and contractility in manipulation tasks. We propose STORM (Slot-based Task-aware Object-centric Representation for robotic Manipulation), a lightweight object-centric adaptation module that augments frozen visual foundation models with a small set of semantic-aware slots for robotic manipulation. Rather than retraining large backbones, STORM employs a multi-phase training strategy: object-centric slots are first stabilized through visual--semantic pretraining using language embeddings, then jointly adapted with a downstream manipulation policy. This staged learning prevents degenerate slot formation and preserves semantic consistency while aligning perception with task objectives. Experiments on object discovery benchmarks and simulated manipulation tasks show that STORM improves generalization to visual distractors, and control performance compared to directly using frozen foundation model features or training object-centric representations end-to-end. Our results highlight multi-phase adaptation as an efficient mechanism for transforming generic foundation model features into task-aware object-centric representations for robotic control.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20381</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 08:46:04 +0000</pubDate>
|
||||
<title>MemOCR: Layout-Aware Visual Memory for Efficient Long-Horizon Reasoning</title>
|
||||
<link>https://arxiv.org/abs/2601.21468</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21468.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yaorui Shi, Shugui Liu, Yu Yang, Wenyu Mao, Yuxin Chen, Qi GU, Hui Su, Xunliang Cai, Xiang Wang, An Zhang</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Long-horizon agentic reasoning necessitates effectively compressing growing interaction histories into a limited context window. Most existing memory systems serialize history as text, where token-level cost is uniform and scales linearly with length, often spending scarce budget on low-value details. To this end, we introduce MemOCR, a multimodal memory agent that improves long-horizon reasoning under tight context budgets by allocating memory space with adaptive information density through visual layout. Concretely, MemOCR maintains a structured rich-text memory (e.g., headings, highlights) and renders it into an image that the agent consults for memory access, visually prioritizing crucial evidence while aggressively compressing auxiliary details. To ensure robustness across varying memory budgets, we train MemOCR with reinforcement learning under budget-aware objectives that expose the agent to diverse compression levels. Across long-context multi-hop and single-hop question-answering benchmarks, MemOCR outperforms strong text-based baselines and achieves more effective context utilization under extreme budgets.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21468</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 09:47:17 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>BMAM: Brain-inspired Multi-Agent Memory Framework</title>
|
||||
<link>https://arxiv.org/abs/2601.20465</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20465.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yang Li, Jiaxiang Liu, Yusong Wang, Yujie Wu, Mingkun Xu</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Language-model-based agents operating over extended interaction horizons face persistent challenges in preserving temporally grounded information and maintaining behavioral consistency across sessions, a failure mode we term soul erosion. We present BMAM (Brain-inspired Multi-Agent Memory), a general-purpose memory architecture that models agent memory as a set of functionally specialized subsystems rather than a single unstructured store. Inspired by cognitive memory systems, BMAM decomposes memory into episodic, semantic, salience-aware, and control-oriented components that operate at complementary time scales. To support long-horizon reasoning, BMAM organizes episodic memories along explicit timelines and retrieves evidence by fusing multiple complementary signals. Experiments on the LoCoMo benchmark show that BMAM achieves 78.45 percent accuracy under the standard long-horizon evaluation setting, and ablation analyses confirm that the hippocampus-inspired episodic memory subsystem plays a critical role in temporal reasoning.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20465</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 10:36:03 +0000</pubDate>
|
||||
<title>LMK > CLS: Landmark Pooling for Dense Embeddings</title>
|
||||
<link>https://arxiv.org/abs/2601.21525</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21525.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Meet Doshi, Aashka Trivedi, Vishwajeet Kumar, Parul Awasthy, Yulong Li, Jaydeep Sen, Radu Florian, Sachindra Joshi</p><p><b>Upvotes:</b> 2</p><p><b>Summary:</b> Representation learning is central to many downstream tasks such as search, clustering, classification, and reranking. State-of-the-art sequence encoders typically collapse a variable-length token sequence to a single vector using a pooling operator, most commonly a special [CLS] token or mean pooling over token embeddings. In this paper, we identify systematic weaknesses of these pooling strategies: [CLS] tends to concentrate information toward the initial positions of the sequence and can under-represent distributed evidence, while mean pooling can dilute salient local signals, sometimes leading to worse short-context performance. To address these issues, we introduce Landmark (LMK) pooling, which partitions a sequence into chunks, inserts landmark tokens between chunks, and forms the final representation by mean-pooling the landmark token embeddings. This simple mechanism improves long-context extrapolation without sacrificing local salient features, at the cost of introducing a small number of special tokens. We empirically demonstrate that LMK pooling matches existing methods on short-context retrieval tasks and yields substantial improvements on long-context tasks, making it a practical and scalable alternative to existing pooling methods.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21525</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 10:40:37 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>AgentLongBench: A Controllable Long Benchmark For Long-Contexts Agents via Environment Rollouts</title>
|
||||
<link>https://arxiv.org/abs/2601.20730</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20730.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shicheng Fang, Yuxin Wang, XiaoRan Liu, Jiahao Lu, Chuanyuan Tan, Xinchi Chen, Yining Zheng. Xuanjing Huang, Xipeng Qiu</p><p><b>Upvotes:</b> 18</p><p><b>Summary:</b> The evolution of Large Language Models (LLMs) into autonomous agents necessitates the management of extensive, dynamic contexts. Current benchmarks, however, remain largely static, relying on passive retrieval tasks that fail to simulate the complexities of agent-environment interaction, such as non-linear reasoning and iterative feedback. To address this, we introduce AgentLongBench, which evaluates agents through simulated environment rollouts based on Lateral Thinking Puzzles. This framework generates rigorous interaction trajectories across knowledge-intensive and knowledge-free scenarios. Experiments with state-of-the-art models and memory systems (32K to 4M tokens) expose a critical weakness: while adept at static retrieval, agents struggle with the dynamic information synthesis essential for workflows. Our analysis indicates that this degradation is driven by the minimum number of tokens required to resolve a query. This factor explains why the high information density inherent in massive tool responses poses a significantly greater challenge than the memory fragmentation typical of long-turn dialogues.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20730</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 16:05:44 +0000</pubDate>
|
||||
<title>KAPSO: A Knowledge-grounded framework for Autonomous Program Synthesis and Optimization</title>
|
||||
<link>https://arxiv.org/abs/2601.21526</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21526.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Alireza Nadaf, Alireza Mohammadshahi, Majid Yazdani</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> We introduce KAPSO, a modular framework for autonomous program synthesis and optimization. Given a natural language goal and an evaluation method, KAPSO iteratively performs ideation, code synthesis and editing, execution, evaluation, and learning to improve a runnable artifact toward measurable objectives. Rather than treating synthesis as the endpoint, KAPSO uses synthesis as an operator within a long-horizon optimization loop, where progress is defined by evaluator outcomes. KAPSO targets long-horizon failures common in coding agents, including lost experimental state, brittle debugging, and weak reuse of domain expertise, by integrating three tightly coupled components. First, a git-native experimentation engine isolates each attempt as a branch, producing reproducible artifacts and preserving provenance across iterations. Second, a knowledge system ingests heterogeneous sources, including repositories, internal playbooks, and curated external resources such as documentation, scientific papers, and web search results, and organizes them into a structured representation that supports retrieval over workflows, implementations, and environment constraints. Third, a cognitive memory layer coordinates retrieval and maintains an episodic store of reusable lessons distilled from experiment traces (run logs, diffs, and evaluator feedback), reducing repeated error modes and accelerating convergence. We evaluated KAPSO on MLE-Bench (Kaggle-style ML competitions) and ALE-Bench (AtCoder heuristic optimization), and report end-to-end performance. Code Available at: https://github.com/Leeroo-AI/kapso</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21526</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 10:40:54 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Idea2Story: An Automated Pipeline for Transforming Research Concepts into Complete Scientific Narratives</title>
|
||||
<link>https://arxiv.org/abs/2601.20833</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20833.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tengyue Xu, Zhuoyang Qian, Gaoge Liu, Li Ling, Zhentao Zhang, Biao Wu, Shuo Zhang, Ke Lu, Wei Shi, Ziqi Wang, Zheng Feng, Yan Luo, Shu Xu, Yongjin Chen, Zhibo Feng, Zhuo Chen, Bruce Yuan, Harry Wang, Kris Chen</p><p><b>Upvotes:</b> 147</p><p><b>Summary:</b> Autonomous scientific discovery with large language model (LLM)-based agents has recently made substantial progress, demonstrating the ability to automate end-to-end research workflows. However, existing systems largely rely on runtime-centric execution paradigms, repeatedly reading, summarizing, and reasoning over large volumes of scientific literature online. This on-the-spot computation strategy incurs high computational cost, suffers from context window limitations, and often leads to brittle reasoning and hallucination. We propose Idea2Story, a pre-computation-driven framework for autonomous scientific discovery that shifts literature understanding from online reasoning to offline knowledge construction. Idea2Story continuously collects peer-reviewed papers together with their review feedback, extracts core methodological units, composes reusable research patterns, and organizes them into a structured methodological knowledge graph. At runtime, underspecified user research intents are aligned to established research paradigms, enabling efficient retrieval and reuse of high-quality research patterns instead of open-ended generation and trial-and-error. By grounding research planning and execution in a pre-built knowledge graph, Idea2Story alleviates the context window bottleneck of LLMs and substantially reduces repeated runtime reasoning over literature. We conduct qualitative analyses and preliminary empirical studies demonstrating that Idea2Story can generate coherent, methodologically grounded, and novel research patterns, and can produce several high-quality research demonstrations in an end-to-end setting. These results suggest that offline knowledge construction provides a practical and scalable foundation for reliable autonomous scientific discovery.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20833</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 18:31:54 +0000</pubDate>
|
||||
<title>ASTRA: Automated Synthesis of agentic Trajectories and Reinforcement Arenas</title>
|
||||
<link>https://arxiv.org/abs/2601.21558</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21558.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiaoyu Tian, Haotian Wang, Shuaiting Chen, Hao Zhou, Kaichi Yu, Yudian Zhang, Jade Ouyang, Junxi Yin, Jiong Chen, Baoyan Guo, Lei Zhang, Junjie Tao, Yuansheng Song, Ming Cui, Chengwei Liu</p><p><b>Upvotes:</b> 45</p><p><b>Summary:</b> Large language models (LLMs) are increasingly used as tool-augmented agents for multi-step decision making, yet training robust tool-using agents remains challenging. Existing methods still require manual intervention, depend on non-verifiable simulated environments, rely exclusively on either supervised fine-tuning (SFT) or reinforcement learning (RL), and struggle with stable long-horizon, multi-turn learning. To address these challenges, we introduce ASTRA, a fully automated end-to-end framework for training tool-augmented language model agents via scalable data synthesis and verifiable reinforcement learning. ASTRA integrates two complementary components. First, a pipeline that leverages the static topology of tool-call graphs synthesizes diverse, structurally grounded trajectories, instilling broad and transferable tool-use competence. Second, an environment synthesis framework that captures the rich, compositional topology of human semantic reasoning converts decomposed question-answer traces into independent, code-executable, and rule-verifiable environments, enabling deterministic multi-turn RL. Based on this method, we develop a unified training methodology that integrates SFT with online RL using trajectory-level rewards to balance task completion and interaction efficiency. Experiments on multiple agentic tool-use benchmarks demonstrate that ASTRA-trained models achieve state-of-the-art performance at comparable scales, approaching closed-source systems while preserving core reasoning ability. We release the full pipelines, environments, and trained models at https://github.com/LianjiaTech/astra.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21558</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 11:22:23 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>DeepSearchQA: Bridging the Comprehensiveness Gap for Deep Research Agents</title>
|
||||
<link>https://arxiv.org/abs/2601.20975</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.20975.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Nikita Gupta, Riju Chatterjee, Lukas Haas, Connie Tao, Andrew Wang, Chang Liu, Hidekazu Oiwa, Elena Gribovskaya, Jan Ackermann, John Blitzer, Sasha Goldshtein, Dipanjan Das</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> We introduce DeepSearchQA, a 900-prompt benchmark for evaluating agents on difficult multi-step information-seeking tasks across 17 different fields. Unlike traditional benchmarks that target single answer retrieval or broad-spectrum factuality, DeepSearchQA features a dataset of challenging, handcrafted tasks designed to evaluate an agent's ability to execute complex search plans to generate exhaustive answer lists. This shift in design explicitly tests three critical, yet under-evaluated capabilities: 1) systematic collation of fragmented information from disparate sources, 2) de-duplication and entity resolution to ensure precision, and 3) the ability to reason about stopping criteria within an open-ended search space. Each task is structured as a causal chain, where discovering information for one step is dependent on the successful completion of the previous one, stressing long-horizon planning and context retention. All tasks are grounded in the open web with objectively verifiable answer sets. Our comprehensive evaluation of state-of-the-art agent architectures reveals significant performance limitations: even the most advanced models struggle to balance high recall with precision. We observe distinct failure modes ranging from premature stopping (under-retrieval) to hedging behaviors, where agents cast an overly wide net of low-confidence answers to artificially boost recall. These findings highlight critical headroom in current agent designs and position DeepSearchQA as an essential diagnostic tool for driving future research toward more robust, deep-research capabilities.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.20975</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 19:20:47 +0000</pubDate>
|
||||
<title>SONIC-O1: A Real-World Benchmark for Evaluating Multimodal Large Language Models on Audio-Video Understanding</title>
|
||||
<link>https://arxiv.org/abs/2601.21666</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21666.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ahmed Y. Radwan, Christos Emmanouilidis, Hina Tabassum, Deval Pandya, Shaina Raza</p><p><b>Upvotes:</b> 0</p><p><b>Summary:</b> Multimodal Large Language Models (MLLMs) are a major focus of recent AI research. However, most prior work focuses on static image understanding, while their ability to process sequential audio-video data remains underexplored. This gap highlights the need for a high-quality benchmark to systematically evaluate MLLM performance in a real-world setting. We introduce SONIC-O1, a comprehensive, fully human-verified benchmark spanning 13 real-world conversational domains with 4,958 annotations and demographic metadata. SONIC-O1 evaluates MLLMs on key tasks, including open-ended summarization, multiple-choice question (MCQ) answering, and temporal localization with supporting rationales (reasoning). Experiments on closed- and open-source models reveal limitations. While the performance gap in MCQ accuracy between two model families is relatively small, we observe a substantial 22.6% performance difference in temporal localization between the best performing closed-source and open-source models. Performance further degrades across demographic groups, indicating persistent disparities in model behavior. Overall, SONIC-O1 provides an open evaluation suite for temporally grounded and socially robust multimodal understanding. We release SONIC-O1 for reproducibility and research: Project page: https://vectorinstitute.github.io/sonic-o1/ Dataset: https://huggingface.co/datasets/vector-institute/sonic-o1 Github: https://github.com/vectorinstitute/sonic-o1 Leaderboard: https://huggingface.co/spaces/vector-institute/sonic-o1-leaderboard</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21666</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 13:01:07 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Llama-3.1-FoundationAI-SecurityLLM-Reasoning-8B Technical Report</title>
|
||||
<link>https://arxiv.org/abs/2601.21051</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21051.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhuoran Yang, Ed Li, Jianliang He, Aman Priyanshu, Baturay Saglam, Paul Kassianik, Sajana Weerawardhena, Anu Vellore, Blaine Nelson, Neusha Javidnia, Arthur Goldblatt, Fraser Burch, Avi Zohary, Assaf Eisenman, Mahdi Sabbaghi, Supriti Vijay, Rahim Dharssi, Dhruv Kedia, Kojin Oshiba, Yaron Singer, Amin Karbasi</p><p><b>Upvotes:</b> 10</p><p><b>Summary:</b> We present Foundation-Sec-8B-Reasoning, the first open-source native reasoning model for cybersecurity. Built upon our previously released Foundation-Sec-8B base model (derived from Llama-3.1-8B-Base), the model is trained through a two-stage process combining supervised fine-tuning (SFT) and reinforcement learning from verifiable rewards (RLVR). Our training leverages proprietary reasoning data spanning cybersecurity analysis, instruction-following, and mathematical reasoning. Evaluation across 10 cybersecurity benchmarks and 10 general-purpose benchmarks demonstrates performance competitive with significantly larger models on cybersecurity tasks while maintaining strong general capabilities. The model shows effective generalization on multi-hop reasoning tasks and strong safety performance when deployed with appropriate system prompts and guardrails. This work demonstrates that domain-specialized reasoning models can achieve strong performance on specialized tasks while maintaining broad general capabilities. We release the model publicly at https://huggingface.co/fdtn-ai/Foundation-Sec-8B-Reasoning.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21051</guid>
|
||||
<pubDate>Wed, 28 Jan 2026 21:15:24 +0000</pubDate>
|
||||
<title>Why Attention Patterns Exist: A Unifying Temporal Perspective Analysis</title>
|
||||
<link>https://arxiv.org/abs/2601.21709</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21709.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Qingyue Yang, Jie Wang, Xing Li, Yinqi Bai, Xialiang Tong, Huiling Zhen, Jianye Hao, Mingxuan Yuan, Bin Li</p><p><b>Upvotes:</b> 2</p><p><b>Summary:</b> Attention patterns play a crucial role in both training and inference of large language models (LLMs). Prior works have identified individual patterns such as retrieval heads, sink heads, and diagonal traces, yet these observations remain fragmented and lack a unifying explanation. To bridge this gap, we introduce Temporal Attention Pattern Predictability Analysis (TAPPA), a unifying framework that explains diverse attention patterns by analyzing their underlying mathematical formulations from a temporally continuous perspective. TAPPA both deepens the understanding of attention behavior and guides inference acceleration approaches. Specifically, TAPPA characterizes attention patterns as predictable patterns with clear regularities and unpredictable patterns that appear effectively random. Our analysis further reveals that this distinction can be explained by the degree of query self-similarity along the temporal dimension. Focusing on the predictable patterns, we further provide a detailed mathematical analysis of three representative cases through the joint effect of queries, keys, and Rotary Positional Embeddings (RoPE). We validate TAPPA by applying its insights to KV cache compression and LLM pruning tasks. Across these tasks, a simple metric motivated by TAPPA consistently improves performance over baseline methods. The code is available at https://github.com/MIRALab-USTC/LLM-TAPPA.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21709</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 13:40:23 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>MAD: Modality-Adaptive Decoding for Mitigating Cross-Modal Hallucinations in Multimodal Large Language Models</title>
|
||||
<link>https://arxiv.org/abs/2601.21181</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21181.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Sangyun Chung, Se Yeon Kim, Youngchae Chee, Yong Man Ro</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Multimodal Large Language Models (MLLMs) suffer from cross-modal hallucinations, where one modality inappropriately influences generation about another, leading to fabricated output. This exposes a more fundamental deficiency in modality-interaction control. To address this, we propose Modality-Adaptive Decoding (MAD), a training-free method that adaptively weights modality-specific decoding branches based on task requirements. MAD leverages the model's inherent ability to self-assess modality relevance by querying which modalities are needed for each task. The extracted modality probabilities are then used to adaptively weight contrastive decoding branches, enabling the model to focus on relevant information while suppressing cross-modal interference. Extensive experiments on CMM and AVHBench demonstrate that MAD significantly reduces cross-modal hallucinations across multiple audio-visual language models (7.8\% and 2.0\% improvements for VideoLLaMA2-AV, 8.7\% and 4.7\% improvements for Qwen2.5-Omni). Our approach demonstrates that explicit modality awareness through self-assessment is crucial for robust multimodal reasoning, offering a principled extension to existing contrastive decoding methods. Our code is available at https://github.com/top-yun/MAD{https://github.com/top-yun/MAD}</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21181</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 02:30:32 +0000</pubDate>
|
||||
<title>DreamActor-M2: Universal Character Image Animation via Spatiotemporal In-Context Learning</title>
|
||||
<link>https://arxiv.org/abs/2601.21716</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21716.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mingshuang Luo, Shuang Liang, Zhengkun Rong, Yuxuan Luo, Tianshu Hu, Ruibing Hou, Hong Chang, Yong Li, Yuan Zhang, Mingyuan Gao</p><p><b>Upvotes:</b> 12</p><p><b>Summary:</b> Character image animation aims to synthesize high-fidelity videos by transferring motion from a driving sequence to a static reference image. Despite recent advancements, existing methods suffer from two fundamental challenges: (1) suboptimal motion injection strategies that lead to a trade-off between identity preservation and motion consistency, manifesting as a "see-saw", and (2) an over-reliance on explicit pose priors (e.g., skeletons), which inadequately capture intricate dynamics and hinder generalization to arbitrary, non-humanoid characters. To address these challenges, we present DreamActor-M2, a universal animation framework that reimagines motion conditioning as an in-context learning problem. Our approach follows a two-stage paradigm. First, we bridge the input modality gap by fusing reference appearance and motion cues into a unified latent space, enabling the model to jointly reason about spatial identity and temporal dynamics by leveraging the generative prior of foundational models. Second, we introduce a self-bootstrapped data synthesis pipeline that curates pseudo cross-identity training pairs, facilitating a seamless transition from pose-dependent control to direct, end-to-end RGB-driven animation. This strategy significantly enhances generalization across diverse characters and motion scenarios. To facilitate comprehensive evaluation, we further introduce AW Bench, a versatile benchmark encompassing a wide spectrum of characters types and motion scenarios. Extensive experiments demonstrate that DreamActor-M2 achieves state-of-the-art performance, delivering superior visual fidelity and robust cross-domain generalization. Project Page: https://grisoon.github.io/DreamActor-M2/</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21716</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 13:43:17 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Scaling Embeddings Outperforms Scaling Experts in Language Models</title>
|
||||
<link>https://arxiv.org/abs/2601.21204</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21204.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Hong Liu, Jiaqi Zhang, Chao Wang, Xing Hu, Linkun Lyu, Jiaqi Sun, Xurui Yang, Bo Wang, Fengcun Li, Yulei Qian, Lingtong Si, Yerui Sun, Rumei Li, Peng Pei, Yuchen Xie, Xunliang Cai</p><p><b>Upvotes:</b> 92</p><p><b>Summary:</b> While Mixture-of-Experts (MoE) architectures have become the standard for sparsity scaling in large language models, they increasingly face diminishing returns and system-level bottlenecks. In this work, we explore embedding scaling as a potent, orthogonal dimension for scaling sparsity. Through a comprehensive analysis and experiments, we identify specific regimes where embedding scaling achieves a superior Pareto frontier compared to expert scaling. We systematically characterize the critical architectural factors governing this efficacy -- ranging from parameter budgeting to the interplay with model width and depth. Moreover, by integrating tailored system optimizations and speculative decoding, we effectively convert this sparsity into tangible inference speedups. Guided by these insights, we introduce LongCat-Flash-Lite, a 68.5B parameter model with ~3B activated trained from scratch. Despite allocating over 30B parameters to embeddings, LongCat-Flash-Lite not only surpasses parameter-equivalent MoE baselines but also exhibits exceptional competitiveness against existing models of comparable scale, particularly in agentic and coding domains.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21204</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 03:11:19 +0000</pubDate>
|
||||
<title>PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing</title>
|
||||
<link>https://arxiv.org/abs/2601.21957</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21957.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Cheng Cui, Ting Sun, Suyin Liang, Tingquan Gao, Zelun Zhang, Jiaxuan Liu, Xueqing Wang, Changda Zhou, Hongen Liu, Manhui Lin, Yue Zhang, Yubo Zhang, Yi Liu, Dianhai Yu, Yanjun Ma</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> We introduce PaddleOCR-VL-1.5, an upgraded model achieving a new state-of-the-art (SOTA) accuracy of 94.5% on OmniDocBench v1.5. To rigorously evaluate robustness against real-world physical distortions, including scanning, skew, warping, screen-photography, and illumination, we propose the Real5-OmniDocBench benchmark. Experimental results demonstrate that this enhanced model attains SOTA performance on the newly curated benchmark. Furthermore, we extend the model's capabilities by incorporating seal recognition and text spotting tasks, while remaining a 0.9B ultra-compact VLM with high efficiency. Code: https://github.com/PaddlePaddle/PaddleOCR</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21957</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 16:35:04 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Reinforcement Learning from Meta-Evaluation: Aligning Language Models Without Ground-Truth Labels</title>
|
||||
<link>https://arxiv.org/abs/2601.21268</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21268.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Micah Rentschler, Jesse Roberts</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> Most reinforcement learning (RL) methods for training large language models (LLMs) require ground-truth labels or task-specific verifiers, limiting scalability when correctness is ambiguous or expensive to obtain. We introduce Reinforcement Learning from Meta-Evaluation (RLME), which optimizes a generator using reward derived from an evaluator's answers to natural-language meta-questions (e.g., "Is the answer correct?" or "Is the reasoning logically consistent?"). RLME treats the evaluator's probability of a positive judgment as a reward and updates the generator via group-relative policy optimization, enabling learning without labels. Across a suite of experiments, we show that RLME achieves accuracy and sample efficiency comparable to label-based training, enables controllable trade-offs among multiple objectives, steers models toward reliable reasoning patterns rather than post-hoc rationalization, and generalizes to open-domain settings where ground-truth labels are unavailable, broadening the domains in which LLMs may be trained with RL.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21268</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 05:02:08 +0000</pubDate>
|
||||
<title>Causal World Modeling for Robot Control</title>
|
||||
<link>https://arxiv.org/abs/2601.21998</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21998.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Lin Li, Qihang Zhang, Yiming Luo, Shuai Yang, Ruilin Wang, Fei Han, Mingrui Yu, Zelin Gao, Nan Xue, Xing Zhu, Yujun Shen, Yinghao Xu</p><p><b>Upvotes:</b> 12</p><p><b>Summary:</b> This work highlights that video world modeling, alongside vision-language pre-training, establishes a fresh and independent foundation for robot learning. Intuitively, video world models provide the ability to imagine the near future by understanding the causality between actions and visual dynamics. Inspired by this, we introduce LingBot-VA, an autoregressive diffusion framework that learns frame prediction and policy execution simultaneously. Our model features three carefully crafted designs: (1) a shared latent space, integrating vision and action tokens, driven by a Mixture-of-Transformers (MoT) architecture, (2) a closed-loop rollout mechanism, allowing for ongoing acquisition of environmental feedback with ground-truth observations, (3) an asynchronous inference pipeline, parallelizing action prediction and motor execution to support efficient control. We evaluate our model on both simulation benchmarks and real-world scenarios, where it shows significant promise in long-horizon manipulation, data efficiency in post-training, and strong generalizability to novel configurations. The code and model are made publicly available to facilitate the community.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21998</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 17:07:43 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>WorldBench: Disambiguating Physics for Diagnostic Evaluation of World Models</title>
|
||||
<link>https://arxiv.org/abs/2601.21282</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21282.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Rishi Upadhyay, Howard Zhang, Jim Solomon, Ayush Agrawal, Pranay Boreddy, Shruti Satya Narayana, Yunhao Ba, Alex Wong, Celso M de Melo, Achuta Kadambi</p><p><b>Upvotes:</b> 0</p><p><b>Summary:</b> Recent advances in generative foundational models, often termed "world models," have propelled interest in applying them to critical tasks like robotic planning and autonomous system training. For reliable deployment, these models must exhibit high physical fidelity, accurately simulating real-world dynamics. Existing physics-based video benchmarks, however, suffer from entanglement, where a single test simultaneously evaluates multiple physical laws and concepts, fundamentally limiting their diagnostic capability. We introduce WorldBench, a novel video-based benchmark specifically designed for concept-specific, disentangled evaluation, allowing us to rigorously isolate and assess understanding of a single physical concept or law at a time. To make WorldBench comprehensive, we design benchmarks at two different levels: 1) an evaluation of intuitive physical understanding with concepts such as object permanence or scale/perspective, and 2) an evaluation of low-level physical constants and material properties such as friction coefficients or fluid viscosity. When SOTA video-based world models are evaluated on WorldBench, we find specific patterns of failure in particular physics concepts, with all tested models lacking the physical consistency required to generate reliable real-world interactions. Through its concept-specific evaluation, WorldBench offers a more nuanced and scalable framework for rigorously evaluating the physical reasoning capabilities of video generation and world models, paving the way for more robust and generalizable world-model-driven learning.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21282</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 05:31:02 +0000</pubDate>
|
||||
<title>Drive-JEPA: Video JEPA Meets Multimodal Trajectory Distillation for End-to-End Driving</title>
|
||||
<link>https://arxiv.org/abs/2601.22032</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22032.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Linhan Wang, Zichong Yang, Chen Bai, Guoxiang Zhang, Xiaotong Liu, Xiaoyin Zheng, Xiao-Xiao Long, Chang-Tien Lu, Cheng Lu</p><p><b>Upvotes:</b> 2</p><p><b>Summary:</b> End-to-end autonomous driving increasingly leverages self-supervised video pretraining to learn transferable planning representations. However, pretraining video world models for scene understanding has so far brought only limited improvements. This limitation is compounded by the inherent ambiguity of driving: each scene typically provides only a single human trajectory, making it difficult to learn multimodal behaviors. In this work, we propose Drive-JEPA, a framework that integrates Video Joint-Embedding Predictive Architecture (V-JEPA) with multimodal trajectory distillation for end-to-end driving. First, we adapt V-JEPA for end-to-end driving, pretraining a ViT encoder on large-scale driving videos to produce predictive representations aligned with trajectory planning. Second, we introduce a proposal-centric planner that distills diverse simulator-generated trajectories alongside human trajectories, with a momentum-aware selection mechanism to promote stable and safe behavior. When evaluated on NAVSIM, the V-JEPA representation combined with a simple transformer-based decoder outperforms prior methods by 3 PDMS in the perception-free setting. The complete Drive-JEPA framework achieves 93.3 PDMS on v1 and 87.8 EPDMS on v2, setting a new state-of-the-art.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22032</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 17:39:20 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Qwen3-ASR Technical Report</title>
|
||||
<link>https://arxiv.org/abs/2601.21337</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21337.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xian Shi, Xiong Wang, Zhifang Guo, Yongqi Wang, Pei Zhang, Xinyu Zhang, Zishan Guo, Hongkun Hao, Yu Xi, Baosong Yang, Jin Xu, Jingren Zhou, Junyang Lin</p><p><b>Upvotes:</b> 22</p><p><b>Summary:</b> In this report, we introduce Qwen3-ASR family, which includes two powerful all-in-one speech recognition models and a novel non-autoregressive speech forced alignment model. Qwen3-ASR-1.7B and Qwen3-ASR-0.6B are ASR models that support language identification and ASR for 52 languages and dialects. Both of them leverage large-scale speech training data and the strong audio understanding ability of their foundation model Qwen3-Omni. We conduct comprehensive internal evaluation besides the open-sourced benchmarks as ASR models might differ little on open-sourced benchmark scores but exhibit significant quality differences in real-world scenarios. The experiments reveal that the 1.7B version achieves SOTA performance among open-sourced ASR models and is competitive with the strongest proprietary APIs while the 0.6B version offers the best accuracy-efficiency trade-off. Qwen3-ASR-0.6B can achieve an average TTFT as low as 92ms and transcribe 2000 seconds speech in 1 second at a concurrency of 128. Qwen3-ForcedAligner-0.6B is an LLM based NAR timestamp predictor that is able to align text-speech pairs in 11 languages. Timestamp accuracy experiments show that the proposed model outperforms the three strongest force alignment models and takes more advantages in efficiency and versatility. To further accelerate the community research of ASR and audio understanding, we release these models under the Apache 2.0 license.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21337</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 06:58:13 +0000</pubDate>
|
||||
<title>Value-Based Pre-Training with Downstream Feedback</title>
|
||||
<link>https://arxiv.org/abs/2601.22108</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22108.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shuqi Ke, Giulia Fanti</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> Can a small amount of verified goal information steer the expensive self-supervised pretraining of foundation models? Standard pretraining optimizes a fixed proxy objective (e.g., next-token prediction), which can misallocate compute away from downstream capabilities of interest. We introduce V-Pretraining: a value-based, modality-agnostic method for controlled continued pretraining in which a lightweight task designer reshapes the pretraining task to maximize the value of each gradient step. For example, consider self-supervised learning (SSL) with sample augmentation. The V-Pretraining task designer selects pretraining tasks (e.g., augmentations) for which the pretraining loss gradient is aligned with a gradient computed over a downstream task (e.g., image segmentation). This helps steer pretraining towards relevant downstream capabilities. Notably, the pretrained model is never updated on downstream task labels; they are used only to shape the pretraining task. Under matched learner update budgets, V-Pretraining of 0.5B--7B language models improves reasoning (GSM8K test Pass@1) by up to 18% relative over standard next-token prediction using only 12% of GSM8K training examples as feedback. In vision SSL, we improve the state-of-the-art results on ADE20K by up to 1.07 mIoU and reduce NYUv2 RMSE while improving ImageNet linear accuracy, and we provide pilot evidence of improved token efficiency in continued pretraining.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22108</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:38:09 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Self-Improving Pretraining: using post-trained models to pretrain better models</title>
|
||||
<link>https://arxiv.org/abs/2601.21343</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21343.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ellen Xiaoqing Tan, Shehzaad Dhuliawala, Jing Xu, Ping Yu, Sainbayar Sukhbaatar, Jason Weston, Olga Golovneva</p><p><b>Upvotes:</b> 10</p><p><b>Summary:</b> Ensuring safety, factuality and overall quality in the generations of large language models is a critical challenge, especially as these models are increasingly deployed in real-world applications. The prevailing approach to addressing these issues involves collecting expensive, carefully curated datasets and applying multiple stages of fine-tuning and alignment. However, even this complex pipeline cannot guarantee the correction of patterns learned during pretraining. Therefore, addressing these issues during pretraining is crucial, as it shapes a model's core behaviors and prevents unsafe or hallucinated outputs from becoming deeply embedded. To tackle this issue, we introduce a new pretraining method that streams documents and uses reinforcement learning (RL) to improve the next K generated tokens at each step. A strong, post-trained model judges candidate generations -- including model rollouts, the original suffix, and a rewritten suffix -- for quality, safety, and factuality. Early in training, the process relies on the original and rewritten suffixes; as the model improves, RL rewards high-quality rollouts. This approach builds higher quality, safer, and more factual models from the ground up. In experiments, our method gives 36.2% and 18.5% relative improvements over standard pretraining in terms of factuality and safety, and up to 86.3% win rate improvements in overall generation quality.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21343</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 07:09:30 +0000</pubDate>
|
||||
<title>Routing the Lottery: Adaptive Subnetworks for Heterogeneous Data</title>
|
||||
<link>https://arxiv.org/abs/2601.22141</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22141.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Grzegorz Stefanski, Alberto Presta, Michal Byra</p><p><b>Upvotes:</b> 2</p><p><b>Summary:</b> In pruning, the Lottery Ticket Hypothesis posits that large networks contain sparse subnetworks, or winning tickets, that can be trained in isolation to match the performance of their dense counterparts. However, most existing approaches assume a single universal winning ticket shared across all inputs, ignoring the inherent heterogeneity of real-world data. In this work, we propose Routing the Lottery (RTL), an adaptive pruning framework that discovers multiple specialized subnetworks, called adaptive tickets, each tailored to a class, semantic cluster, or environmental condition. Across diverse datasets and tasks, RTL consistently outperforms single- and multi-model baselines in balanced accuracy and recall, while using up to 10 times fewer parameters than independent models and exhibiting semantically aligned. Furthermore, we identify subnetwork collapse, a performance drop under aggressive pruning, and introduce a subnetwork similarity score that enables label-free diagnosis of oversparsification. Overall, our results recast pruning as a mechanism for aligning model structure with data heterogeneity, paving the way toward more modular and context-aware deep learning.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22141</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:56:41 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Generation Enhances Understanding in Unified Multimodal Models via Multi-Representation Generation</title>
|
||||
<link>https://arxiv.org/abs/2601.21406</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21406.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zihan Su, Hongyang Wei, Kangrui Cen, Yong Wang, Guanhua Chen, Chun Yuan, Xiangxiang Chu</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Unified Multimodal Models (UMMs) integrate both visual understanding and generation within a single framework. Their ultimate aspiration is to create a cycle where understanding and generation mutually reinforce each other. While recent post-training methods have successfully leveraged understanding to enhance generation, the reverse direction of utilizing generation to improve understanding remains largely unexplored. In this work, we propose UniMRG (Unified Multi-Representation Generation), a simple yet effective architecture-agnostic post-training method. UniMRG enhances the understanding capabilities of UMMs by incorporating auxiliary generation tasks. Specifically, we train UMMs to generate multiple intrinsic representations of input images, namely pixel (reconstruction), depth (geometry), and segmentation (structure), alongside standard visual understanding objectives. By synthesizing these diverse representations, UMMs capture complementary information regarding appearance, spatial relations, and structural layout. Consequently, UMMs develop a deeper and more comprehensive understanding of visual inputs. Extensive experiments across diverse UMM architectures demonstrate that our method notably enhances fine-grained perception, reduces hallucinations, and improves spatial understanding, while simultaneously boosting generation capabilities.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21406</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 08:42:25 +0000</pubDate>
|
||||
<title>SSL: Sweet Spot Learning for Differentiated Guidance in Agentic Optimization</title>
|
||||
<link>https://arxiv.org/abs/2601.22491</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22491.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jinyang Wu, Changpeng Yang, Yuhao Shen, Fangzhi Xu, Bolin Ni, Chonghua Liao, Yuchen Liu, Hongzhen Wang, Shuai Nie, Shuai Zhang, Haoran Luo, Jiaming Xu</p><p><b>Upvotes:</b> 11</p><p><b>Summary:</b> Reinforcement learning with verifiable rewards has emerged as a powerful paradigm for training intelligent agents. However, existing methods typically employ binary rewards that fail to capture quality differences among trajectories achieving identical outcomes, thereby overlooking potential diversity within the solution space. Inspired by the ``sweet spot'' concept in tennis-the racket's core region that produces optimal hitting effects, we introduce Sweet Spot Learning (SSL), a novel framework that provides differentiated guidance for agent optimization. SSL follows a simple yet effective principle: progressively amplified, tiered rewards guide policies toward the sweet-spot region of the solution space. This principle naturally adapts across diverse tasks: visual perception tasks leverage distance-tiered modeling to reward proximity, while complex reasoning tasks reward incremental progress toward promising solutions. We theoretically demonstrate that SSL preserves optimal solution ordering and enhances the gradient signal-to-noise ratio, thereby fostering more directed optimization. Extensive experiments across GUI perception, short/long-term planning, and complex reasoning tasks show consistent improvements over strong baselines on 12 benchmarks, achieving up to 2.5X sample efficiency gains and effective cross-task transferability. Our work establishes SSL as a general principle for training capable and robust agents.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22491</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 03:02:18 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Spotlighting Task-Relevant Features: Object-Centric Representations for Better Generalization in Robotic Manipulation</title>
|
||||
<link>https://arxiv.org/abs/2601.21416</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21416.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Alexandre Chapin, Bruno Machado, Emmanuel Dellandréa, Liming Chen</p><p><b>Upvotes:</b> 0</p><p><b>Summary:</b> The generalization capabilities of robotic manipulation policies are heavily influenced by the choice of visual representations. Existing approaches typically rely on representations extracted from pre-trained encoders, using two dominant types of features: global features, which summarize an entire image via a single pooled vector, and dense features, which preserve a patch-wise embedding from the final encoder layer. While widely used, both feature types mix task-relevant and irrelevant information, leading to poor generalization under distribution shifts, such as changes in lighting, textures, or the presence of distractors. In this work, we explore an intermediate structured alternative: Slot-Based Object-Centric Representations (SBOCR), which group dense features into a finite set of object-like entities. This representation permits to naturally reduce the noise provided to the robotic manipulation policy while keeping enough information to efficiently perform the task. We benchmark a range of global and dense representations against intermediate slot-based representations, across a suite of simulated and real-world manipulation tasks ranging from simple to complex. We evaluate their generalization under diverse visual conditions, including changes in lighting, texture, and the presence of distractors. Our findings reveal that SBOCR-based policies outperform dense and global representation-based policies in generalization settings, even without task-specific pretraining. These insights suggest that SBOCR is a promising direction for designing visual systems that generalize effectively in dynamic, real-world robotic environments.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21416</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 08:55:53 +0000</pubDate>
|
||||
<title>TTCS: Test-Time Curriculum Synthesis for Self-Evolving</title>
|
||||
<link>https://arxiv.org/abs/2601.22628</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22628.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chengyi Yang, Zhishang Xiang, Yunbo Tang, Zongpei Teng, Chengsong Huang, Fei Long, Yuhan Liu, Jinsong Su</p><p><b>Upvotes:</b> 24</p><p><b>Summary:</b> Test-Time Training offers a promising way to improve the reasoning ability of large language models (LLMs) by adapting the model using only the test questions. However, existing methods struggle with difficult reasoning problems for two reasons: raw test questions are often too difficult to yield high-quality pseudo-labels, and the limited size of test sets makes continuous online updates prone to instability. To address these limitations, we propose TTCS, a co-evolving test-time training framework. Specifically, TTCS initializes two policies from the same pretrained model: a question synthesizer and a reasoning solver. These policies evolve through iterative optimization: the synthesizer generates progressively challenging question variants conditioned on the test questions, creating a structured curriculum tailored to the solver's current capability, while the solver updates itself using self-consistency rewards computed from multiple sampled responses on both original test and synthetic questions. Crucially, the solver's feedback guides the synthesizer to generate questions aligned with the model's current capability, and the generated question variants in turn stabilize the solver's test-time training. Experiments show that TTCS consistently strengthens the reasoning ability on challenging mathematical benchmarks and transfers to general-domain tasks across different LLM backbones, highlighting a scalable path towards dynamically constructing test-time curricula for self-evolving. Our code and implementation details are available at https://github.com/XMUDeepLIT/TTCS.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22628</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 06:38:02 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>ConceptMoE: Adaptive Token-to-Concept Compression for Implicit Compute Allocation</title>
|
||||
<link>https://arxiv.org/abs/2601.21420</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21420.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zihao Huang, Jundong Zhou, Xingwei Qu, Qiyang Min, Ge Zhang</p><p><b>Upvotes:</b> 33</p><p><b>Summary:</b> Large language models allocate uniform computation across all tokens, ignoring that some sequences are trivially predictable while others require deep reasoning. We introduce ConceptMoE, which dynamically merges semantically similar tokens into concept representations, performing implicit token-level compute allocation. A learnable chunk module identifies optimal boundaries by measuring inter-token similarity, compressing sequences by a target ratio R before they enter the compute-intensive concept model. Crucially, the MoE architecture enables controlled evaluation: we reallocate saved computation to match baseline activated FLOPs (excluding attention map computation) and total parameters, isolating genuine architectural benefits. Under these conditions, ConceptMoE consistently outperforms standard MoE across language and vision-language tasks, achieving +0.9 points on language pretraining, +2.3 points on long context understanding, and +0.6 points on multimodal benchmarks. When converting pretrained MoE during continual training with layer looping, gains reach +5.5 points, demonstrating practical applicability. Beyond performance, ConceptMoE reduces attention computation by up to R^2times and KV cache by Rtimes. At R=2, empirical measurements show prefill speedups reaching 175\% and decoding speedups up to 117\% on long sequences. The minimal architectural modifications enable straightforward integration into existing MoE, demonstrating that adaptive concept-level processing fundamentally improves both effectiveness and efficiency of large language models.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21420</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 08:58:22 +0000</pubDate>
|
||||
<title>Statistical Estimation of Adversarial Risk in Large Language Models under Best-of-N Sampling</title>
|
||||
<link>https://arxiv.org/abs/2601.22636</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22636.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mingqian Feng, Xiaodong Liu, Weiwei Yang, Chenliang Xu, Christopher White, Jianfeng Gao</p><p><b>Upvotes:</b> 8</p><p><b>Summary:</b> Large Language Models (LLMs) are typically evaluated for safety under single-shot or low-budget adversarial prompting, which underestimates real-world risk. In practice, attackers can exploit large-scale parallel sampling to repeatedly probe a model until a harmful response is produced. While recent work shows that attack success increases with repeated sampling, principled methods for predicting large-scale adversarial risk remain limited. We propose a scaling-aware Best-of-N estimation of risk, SABER, for modeling jailbreak vulnerability under Best-of-N sampling. We model sample-level success probabilities using a Beta distribution, the conjugate prior of the Bernoulli distribution, and derive an analytic scaling law that enables reliable extrapolation of large-N attack success rates from small-budget measurements. Using only n=100 samples, our anchored estimator predicts ASR@1000 with a mean absolute error of 1.66, compared to 12.04 for the baseline, which is an 86.2% reduction in estimation error. Our results reveal heterogeneous risk scaling profiles and show that models appearing robust under standard evaluation can experience rapid nonlinear risk amplification under parallel adversarial pressure. This work provides a low-cost, scalable methodology for realistic LLM safety assessment. We will release our code and evaluation scripts upon publication to future research.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22636</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 06:54:35 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Shaping capabilities with token-level data filtering</title>
|
||||
<link>https://arxiv.org/abs/2601.21571</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21571.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Neil Rathi, Alec Radford</p><p><b>Upvotes:</b> 17</p><p><b>Summary:</b> Current approaches to reducing undesired capabilities in language models are largely post hoc, and can thus be easily bypassed by adversaries. A natural alternative is to shape capabilities during pretraining itself. On the proxy task of removing medical capabilities, we show that the simple intervention of filtering pretraining data is highly effective, robust, and inexpensive at scale. Inspired by work on data attribution, we show that filtering tokens is more effective than filtering documents, achieving the same hit to undesired capabilities at a lower cost to benign ones. Training models spanning two orders of magnitude, we then demonstrate that filtering gets more effective with scale: for our largest models, token filtering leads to a 7000x compute slowdown on the forget domain. We also show that models trained with token filtering can still be aligned on the forget domain. Along the way, we introduce a methodology for labeling tokens with sparse autoencoders and distilling cheap, high-quality classifiers. We also demonstrate that filtering can be robust to noisy labels with sufficient pretraining compute.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21571</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 11:34:01 +0000</pubDate>
|
||||
<title>Pushing the Boundaries of Natural Reasoning: Interleaved Bonus from Formal-Logic Verification</title>
|
||||
<link>https://arxiv.org/abs/2601.22642</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22642.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chuxue Cao, Jinluan Yang, Haoran Li, Kunhao Pan, Zijian Zhao, Zhengyu Chen, Yuchen Tian, Lijun Wu, Conghui He, Sirui Han, Yike Guo</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Large Language Models (LLMs) show remarkable capabilities, yet their stochastic next-token prediction creates logical inconsistencies and reward hacking that formal symbolic systems avoid. To bridge this gap, we introduce a formal logic verification-guided framework that dynamically interleaves formal symbolic verification with the natural language generation process, providing real-time feedback to detect and rectify errors as they occur. Distinguished from previous neuro-symbolic methods limited by passive post-hoc validation, our approach actively penalizes intermediate fallacies during the reasoning chain. We operationalize this framework via a novel two-stage training pipeline that synergizes formal logic verification-guided supervised fine-tuning and policy optimization. Extensive evaluation on six benchmarks spanning mathematical, logical, and general reasoning demonstrates that our 7B and 14B models outperform state-of-the-art baselines by average margins of 10.4% and 14.2%, respectively. These results validate that formal verification can serve as a scalable mechanism to significantly push the performance boundaries of advanced LLM reasoning.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22642</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 07:01:25 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>KromHC: Manifold-Constrained Hyper-Connections with Kronecker-Product Residual Matrices</title>
|
||||
<link>https://arxiv.org/abs/2601.21579</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21579.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Wuyang Zhou, Yuxuan Gu, Giorgos Iacovides, Danilo Mandic</p><p><b>Upvotes:</b> 5</p><p><b>Summary:</b> The success of Hyper-Connections (HC) in neural networks (NN) has also highlighted issues related to its training instability and restricted scalability. The Manifold-Constrained Hyper-Connections (mHC) mitigate these challenges by projecting the residual connection space onto a Birkhoff polytope, however, it faces two issues: 1) its iterative Sinkhorn-Knopp (SK) algorithm does not always yield exact doubly stochastic residual matrices; 2) mHC incurs a prohibitive O(n^3C) parameter complexity with n as the width of the residual stream and C as the feature dimension. The recently proposed mHC-lite reparametrizes the residual matrix via the Birkhoff-von-Neumann theorem to guarantee double stochasticity, but also faces a factorial explosion in its parameter complexity, O left( nC cdot n! right). To address both challenges, we propose KromHC, which uses the Kronecker products of smaller doubly stochastic matrices to parametrize the residual matrix in mHC. By enforcing manifold constraints across the factor residual matrices along each mode of the tensorized residual stream, KromHC guarantees exact double stochasticity of the residual matrices while reducing parameter complexity to O(n^2C). Comprehensive experiments demonstrate that KromHC matches or even outperforms state-of-the-art (SOTA) mHC variants, while requiring significantly fewer trainable parameters. The code is available at https://github.com/wz1119/KromHC.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21579</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 11:43:05 +0000</pubDate>
|
||||
<title>Real-Time Aligned Reward Model beyond Semantics</title>
|
||||
<link>https://arxiv.org/abs/2601.22664</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22664.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zixuan Huang, Xin Xia, Yuxi Ren, Jianbin Zheng, Xuefeng Xiao, Hongyan Xie, Li Huaqiu, Songshi Liang, Zhongxiang Dai, Fuzhen Zhuang, Jianxin Li, Yikun Ban, Deqing Wang</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Reinforcement Learning from Human Feedback (RLHF) is a pivotal technique for aligning large language models (LLMs) with human preferences, yet it is susceptible to reward overoptimization, in which policy models overfit to the reward model, exploit spurious reward patterns instead of faithfully capturing human intent. Prior mitigations primarily relies on surface semantic information and fails to efficiently address the misalignment between the reward model (RM) and the policy model caused by continuous policy distribution shifts. This inevitably leads to an increasing reward discrepancy, exacerbating reward overoptimization. To address these limitations, we introduce R2M (Real-Time Aligned Reward Model), a novel lightweight RLHF framework. R2M goes beyond vanilla reward models that solely depend on the semantic representations of a pretrained LLM. Instead, it leverages the evolving hidden states of the policy (namely policy feedback) to align with the real-time distribution shift of the policy during the RL process. This work points to a promising new direction for improving the performance of reward models through real-time utilization of feedback from policy models.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22664</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 07:32:35 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening</title>
|
||||
<link>https://arxiv.org/abs/2601.21590</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21590.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiaotong Ji, Rasul Tutunov, Matthieu Zimmer, Haitham Bou Ammar</p><p><b>Upvotes:</b> 11</p><p><b>Summary:</b> Reinforcement learning (RL) post-training is a dominant approach for improving the reasoning performance of large language models (LLMs), yet growing evidence suggests that its gains arise primarily from distribution sharpening rather than the acquisition of new capabilities. Recent work has shown that sampling from the power distribution of LLMs using Markov chain Monte Carlo (MCMC) can recover performance comparable to RL post-training without relying on external rewards; however, the high computational cost of MCMC makes such approaches impractical for widespread adoption. In this work, we propose a theoretically grounded alternative that eliminates the need for iterative MCMC. We derive a novel formulation showing that the global power distribution can be approximated by a token-level scaled low-temperature one, where the scaling factor captures future trajectory quality. Leveraging this insight, we introduce a training-free and verifier-free algorithm that sharpens the base model's generative distribution autoregressively. Empirically, we evaluate our method on math, QA, and code tasks across four LLMs, and show that our method matches or surpasses one-shot GRPO without relying on any external rewards, while reducing inference latency by over 10x compared to MCMC-based sampling.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21590</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 12:01:53 +0000</pubDate>
|
||||
<title>ExpAlign: Expectation-Guided Vision-Language Alignment for Open-Vocabulary Grounding</title>
|
||||
<link>https://arxiv.org/abs/2601.22666</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22666.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junyi Hu, Tian Bai, Fengyi Wu, Wenyan Li, Zhenming Peng, Yi Zhang</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> Open-vocabulary grounding requires accurate vision-language alignment under weak supervision, yet existing methods either rely on global sentence embeddings that lack fine-grained expressiveness or introduce token-level alignment with explicit supervision or heavy cross-attention designs. We propose ExpAlign, a theoretically grounded vision-language alignment framework built on a principled multiple instance learning formulation. ExpAlign introduces an Expectation Alignment Head that performs attention-based soft MIL pooling over token-region similarities, enabling implicit token and instance selection without additional annotations. To further stabilize alignment learning, we develop an energy-based multi-scale consistency regularization scheme, including a Top-K multi-positive contrastive objective and a Geometry-Aware Consistency Objective derived from a Lagrangian-constrained free-energy minimization. Extensive experiments show that ExpAlign consistently improves open-vocabulary detection and zero-shot instance segmentation, particularly on long-tail categories. Most notably, it achieves 36.2 AP_r on the LVIS minival split, outperforming other state-of-the-art methods at comparable model scale, while remaining lightweight and inference-efficient.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22666</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 07:38:04 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Beyond Imitation: Reinforcement Learning for Active Latent Planning</title>
|
||||
<link>https://arxiv.org/abs/2601.21598</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21598.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhi Zheng, Wee Sun Lee</p><p><b>Upvotes:</b> 6</p><p><b>Summary:</b> Aiming at efficient and dense chain-of-thought (CoT) reasoning, latent reasoning methods fine-tune Large Language Models (LLMs) to substitute discrete language tokens with continuous latent tokens. These methods consume fewer tokens compared to the conventional language CoT reasoning and have the potential to plan in a dense latent space. However, current latent tokens are generally supervised based on imitating language labels. Considering that there can be multiple equivalent but diverse CoT labels for a question, passively imitating an arbitrary one may lead to inferior latent token representations and latent reasoning policies, undermining the potential planning ability and resulting in clear gaps between training and testing. In this work, we emphasize the importance of active planning over the representation space of latent tokens in achieving the optimal latent reasoning policy. So, we propose the Active Latent Planning method (ATP-Latent), which models the supervision process of latent tokens as a conditional variational auto-encoder (VAE) to obtain a smoother latent space. Moreover, to facilitate the most reasonable latent reasoning policy, ATP-Latent conducts reinforcement learning (RL) with an auxiliary coherence reward, which is calculated based on the consistency between VAE-decoded contents of latent tokens, enabling a guided RL process. In experiments on LLaMA-1B, ATP-Latent demonstrates +4.1\% accuracy and -3.3\% tokens on four benchmarks compared to advanced baselines. Codes are available on https://github.com/zz1358m/ATP-Latent-master.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21598</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 12:07:16 +0000</pubDate>
|
||||
<title>Visual Personalization Turing Test</title>
|
||||
<link>https://arxiv.org/abs/2601.22680</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22680.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Rameen Abdal, James Burgess, Sergey Tulyakov, Kuan-Chieh Jackson Wang</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> We introduce the Visual Personalization Turing Test (VPTT), a new paradigm for evaluating contextual visual personalization based on perceptual indistinguishability, rather than identity replication. A model passes the VPTT if its output (image, video, 3D asset, etc.) is indistinguishable to a human or calibrated VLM judge from content a given person might plausibly create or share. To operationalize VPTT, we present the VPTT Framework, integrating a 10k-persona benchmark (VPTT-Bench), a visual retrieval-augmented generator (VPRAG), and the VPTT Score, a text-only metric calibrated against human and VLM judgments. We show high correlation across human, VLM, and VPTT evaluations, validating the VPTT Score as a reliable perceptual proxy. Experiments demonstrate that VPRAG achieves the best alignment-originality balance, offering a scalable and privacy-safe foundation for personalized generative AI.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22680</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 07:53:07 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>OCRVerse: Towards Holistic OCR in End-to-End Vision-Language Models</title>
|
||||
<link>https://arxiv.org/abs/2601.21639</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21639.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yufeng Zhong, Lei Chen, Xuanle Zhao, Wenkang Han, Liming Zheng, Jing Huang, Deyang Jiang, Yilin Cao, Lin Ma, Zhixiong Zeng</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> The development of large vision language models drives the demand for managing, and applying massive amounts of multimodal data, making OCR technology, which extracts information from visual images, increasingly popular. However, existing OCR methods primarily focus on recognizing text elements from images or scanned documents (Text-centric OCR), neglecting the identification of visual elements from visually information-dense image sources (Vision-centric OCR), such as charts, web pages and science plots. In reality, these visually information-dense images are widespread on the internet and have significant real-world application value, such as data visualization and web page analysis. In this technical report, we propose OCRVerse, the first holistic OCR method in end-to-end manner that enables unified text-centric OCR and vision-centric OCR. To this end, we constructe comprehensive data engineering to cover a wide range of text-centric documents, such as newspapers, magazines and books, as well as vision-centric rendered composites, including charts, web pages and scientific plots. Moreover, we propose a two-stage SFT-RL multi-domain training method for OCRVerse. SFT directly mixes cross-domain data to train and establish initial domain knowledge, while RL focuses on designing personalized reward strategies for the characteristics of each domain. Specifically, since different domains require various output formats and expected outputs, we provide sufficient flexibility in the RL stage to customize flexible reward signals for each domain, thereby improving cross-domain fusion and avoiding data conflicts. Experimental results demonstrate the effectiveness of OCRVerse, achieving competitive results across text-centric and vision-centric data types, even comparable to large-scale open-source and closed-source models.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21639</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 12:43:02 +0000</pubDate>
|
||||
<title>Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation</title>
|
||||
<link>https://arxiv.org/abs/2601.22813</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22813.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Andrei Panferov, Erik Schultheis, Soroush Tabesh, Dan Alistarh</p><p><b>Upvotes:</b> 40</p><p><b>Summary:</b> The NVFP4 lower-precision format, supported in hardware by NVIDIA Blackwell GPUs, promises to allow, for the first time, end-to-end fully-quantized pre-training of massive models such as LLMs. Yet, existing quantized training methods still sacrifice some of the representation capacity of this format in favor of more accurate unbiased quantized gradient estimation by stochastic rounding (SR), losing noticeable accuracy relative to standard FP16 and FP8 training. In this paper, improve the state of the art for quantized training in NVFP4 via a novel unbiased quantization routine for micro-scaled formats, called MS-EDEN, that has more than 2x lower quantization error than SR. We integrate it into a novel fully-NVFP4 quantization scheme for linear layers, called Quartet II. We show analytically that Quartet II achieves consistently better gradient estimation across all major matrix multiplications, both on the forward and on the backward passes. In addition, our proposal synergizes well with recent training improvements aimed specifically at NVFP4. We further validate Quartet II on end-to-end LLM training with up to 1.9B parameters on 38B tokens. We provide kernels for execution on NVIDIA Blackwell GPUs with up to 4.2x speedup over BF16. Our code is available at https://github.com/IST-DASLab/Quartet-II .</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22813</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 10:39:11 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Language-based Trial and Error Falls Behind in the Era of Experience</title>
|
||||
<link>https://arxiv.org/abs/2601.21754</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21754.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haoyu Wang, Guozheng Ma, Shugang Cui, Yilun Kong, Haotian Luo, Li Shen, Mengya Gao, Yichao Wu, Xiaogang Wang, Dacheng Tao</p><p><b>Upvotes:</b> 15</p><p><b>Summary:</b> While Large Language Models (LLMs) excel in language-based agentic tasks, their applicability to unseen, nonlinguistic environments (e.g., symbolic or spatial tasks) remains limited. Previous work attributes this performance gap to the mismatch between the pretraining distribution and the testing distribution. In this work, we demonstrate the primary bottleneck is the prohibitive cost of exploration: mastering these tasks requires extensive trial-and-error, which is computationally unsustainable for parameter-heavy LLMs operating in a high dimensional semantic space. To address this, we propose SCOUT (Sub-Scale Collaboration On Unseen Tasks), a novel framework that decouples exploration from exploitation. We employ lightweight "scouts" (e.g., small MLPs) to probe environmental dynamics at a speed and scale far exceeding LLMs. The collected trajectories are utilized to bootstrap the LLM via Supervised Fine-Tuning (SFT), followed by multi-turn Reinforcement Learning (RL) to activate its latent world knowledge. Empirically, SCOUT enables a Qwen2.5-3B-Instruct model to achieve an average score of 0.86, significantly outperforming proprietary models, including Gemini-2.5-Pro (0.60), while saving about 60% GPU hours consumption.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21754</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 14:08:41 +0000</pubDate>
|
||||
<title>NativeTok: Native Visual Tokenization for Improved Image Generation</title>
|
||||
<link>https://arxiv.org/abs/2601.22837</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22837.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Bin Wu, Mengqi Huang, Weinan Jia, Zhendong Mao</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> VQ-based image generation typically follows a two-stage pipeline: a tokenizer encodes images into discrete tokens, and a generative model learns their dependencies for reconstruction. However, improved tokenization in the first stage does not necessarily enhance the second-stage generation, as existing methods fail to constrain token dependencies. This mismatch forces the generative model to learn from unordered distributions, leading to bias and weak coherence. To address this, we propose native visual tokenization, which enforces causal dependencies during tokenization. Building on this idea, we introduce NativeTok, a framework that achieves efficient reconstruction while embedding relational constraints within token sequences. NativeTok consists of: (1) a Meta Image Transformer (MIT) for latent image modeling, and (2) a Mixture of Causal Expert Transformer (MoCET), where each lightweight expert block generates a single token conditioned on prior tokens and latent features. We further design a Hierarchical Native Training strategy that updates only new expert blocks, ensuring training efficiency. Extensive experiments demonstrate the effectiveness of NativeTok.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22837</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 11:01:43 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>MMFineReason: Closing the Multimodal Reasoning Gap via Open Data-Centric Methods</title>
|
||||
<link>https://arxiv.org/abs/2601.21821</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21821.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Honglin Lin, Zheng Liu, Yun Zhu, Chonghan Qin, Juekai Lin, Xiaoran Shang, Conghui He, Wentao Zhang, Lijun Wu</p><p><b>Upvotes:</b> 47</p><p><b>Summary:</b> Recent advances in Vision Language Models (VLMs) have driven significant progress in visual reasoning. However, open-source VLMs still lag behind proprietary systems, largely due to the lack of high-quality reasoning data. Existing datasets offer limited coverage of challenging domains such as STEM diagrams and visual puzzles, and lack consistent, long-form Chain-of-Thought (CoT) annotations essential for eliciting strong reasoning capabilities. To bridge this gap, we introduce MMFineReason, a large-scale multimodal reasoning dataset comprising 1.8M samples and 5.1B solution tokens, featuring high-quality reasoning annotations distilled from Qwen3-VL-235B-A22B-Thinking. The dataset is established via a systematic three-stage pipeline: (1) large-scale data collection and standardization, (2) CoT rationale generation, and (3) comprehensive selection based on reasoning quality and difficulty awareness. The resulting dataset spans STEM problems, visual puzzles, games, and complex diagrams, with each sample annotated with visually grounded reasoning traces. We fine-tune Qwen3-VL-Instruct on MMFineReason to develop MMFineReason-2B/4B/8B versions. Our models establish new state-of-the-art results for their size class. Notably, MMFineReason-4B succesfully surpasses Qwen3-VL-8B-Thinking, and MMFineReason-8B even outperforms Qwen3-VL-30B-A3B-Thinking while approaching Qwen3-VL-32B-Thinking, demonstrating remarkable parameter efficiency. Crucially, we uncover a "less is more" phenomenon via our difficulty-aware filtering strategy: a subset of just 7\% (123K samples) achieves performance comparable to the full dataset. Notably, we reveal a synergistic effect where reasoning-oriented data composition simultaneously boosts general capabilities.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21821</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 15:07:28 +0000</pubDate>
|
||||
<title>DINO-SAE: DINO Spherical Autoencoder for High-Fidelity Image Reconstruction and Generation</title>
|
||||
<link>https://arxiv.org/abs/2601.22904</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22904.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Hun Chang, Byunghee Cha, Jong Chul Ye</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Recent studies have explored using pretrained Vision Foundation Models (VFMs) such as DINO for generative autoencoders, showing strong generative performance. Unfortunately, existing approaches often suffer from limited reconstruction fidelity due to the loss of high-frequency details. In this work, we present the DINO Spherical Autoencoder (DINO-SAE), a framework that bridges semantic representation and pixel-level reconstruction. Our key insight is that semantic information in contrastive representations is primarily encoded in the direction of feature vectors, while forcing strict magnitude matching can hinder the encoder from preserving fine-grained details. To address this, we introduce Hierarchical Convolutional Patch Embedding module that enhances local structure and texture preservation, and Cosine Similarity Alignment objective that enforces semantic consistency while allowing flexible feature magnitudes for detail retention. Furthermore, leveraging the observation that SSL-based foundation model representations intrinsically lie on a hypersphere, we employ Riemannian Flow Matching to train a Diffusion Transformer (DiT) directly on this spherical latent manifold. Experiments on ImageNet-1K demonstrate that our approach achieves state-of-the-art reconstruction quality, reaching 0.37 rFID and 26.2 dB PSNR, while maintaining strong semantic alignment to the pretrained VFM. Notably, our Riemannian Flow Matching-based DiT exhibits efficient convergence, achieving a gFID of 3.47 at 80 epochs.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22904</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 12:25:34 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>WebArbiter: A Principle-Guided Reasoning Process Reward Model for Web Agents</title>
|
||||
<link>https://arxiv.org/abs/2601.21872</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21872.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yao Zhang, Shijie Tang, Zeyu Li, Zhen Han, Volker Tresp</p><p><b>Upvotes:</b> 0</p><p><b>Summary:</b> Web agents hold great potential for automating complex computer tasks, yet their interactions involve long-horizon, sequential decision-making with irreversible actions. In such settings, outcome-based supervision is sparse and delayed, often rewarding incorrect trajectories and failing to support inference-time scaling. This motivates the use of Process Reward Models (WebPRMs) for web navigation, but existing approaches remain limited: scalar WebPRMs collapse progress into coarse, weakly grounded signals, while checklist-based WebPRMs rely on brittle template matching that fails under layout or semantic changes and often mislabels superficially correct actions as successful, providing little insight or interpretability. To address these challenges, we introduce WebArbiter, a reasoning-first, principle-inducing WebPRM that formulates reward modeling as text generation, producing structured justifications that conclude with a preference verdict and identify the action most conducive to task completion under the current context. Training follows a two-stage pipeline: reasoning distillation equips the model with coherent principle-guided reasoning, and reinforcement learning corrects teacher biases by directly aligning verdicts with correctness, enabling stronger generalization. To support systematic evaluation, we release WebPRMBench, a comprehensive benchmark spanning four diverse web environments with rich tasks and high-quality preference annotations. On WebPRMBench, WebArbiter-7B outperforms the strongest baseline, GPT-5, by 9.1 points. In reward-guided trajectory search on WebArena-Lite, it surpasses the best prior WebPRM by up to 7.2 points, underscoring its robustness and practical value in real-world complex web tasks.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21872</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 15:39:50 +0000</pubDate>
|
||||
<title>Golden Goose: A Simple Trick to Synthesize Unlimited RLVR Tasks from Unverifiable Internet Text</title>
|
||||
<link>https://arxiv.org/abs/2601.22975</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22975.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ximing Lu, David Acuna, Jaehun Jung, Jian Hu, Di Zhang, Shizhe Diao, Yunheng Zou, Shaokun Zhang, Brandon Cui, Mingjie Liu, Hyunwoo Kim, Prithviraj Ammanabrolu, Jan Kautz, Yi Dong, Yejin Choi</p><p><b>Upvotes:</b> 33</p><p><b>Summary:</b> Reinforcement Learning with Verifiable Rewards (RLVR) has become a cornerstone for unlocking complex reasoning in Large Language Models (LLMs). Yet, scaling up RL is bottlenecked by limited existing verifiable data, where improvements increasingly saturate over prolonged training. To overcome this, we propose Golden Goose, a simple trick to synthesize unlimited RLVR tasks from unverifiable internet text by constructing a multiple-choice question-answering version of the fill-in-the-middle task. Given a source text, we prompt an LLM to identify and mask key reasoning steps, then generate a set of diverse, plausible distractors. This enables us to leverage reasoning-rich unverifiable corpora typically excluded from prior RLVR data construction (e.g., science textbooks) to synthesize GooseReason-0.7M, a large-scale RLVR dataset with over 0.7 million tasks spanning mathematics, programming, and general scientific domains. Empirically, GooseReason effectively revives models saturated on existing RLVR data, yielding robust, sustained gains under continuous RL and achieving new state-of-the-art results for 1.5B and 4B-Instruct models across 15 diverse benchmarks. Finally, we deploy Golden Goose in a real-world setting, synthesizing RLVR tasks from raw FineWeb scrapes for the cybersecurity domain, where no prior RLVR data exists. Training Qwen3-4B-Instruct on the resulting data GooseReason-Cyber sets a new state-of-the-art in cybersecurity, surpassing a 7B domain-specialized model with extensive domain-specific pre-training and post-training. This highlights the potential of automatically scaling up RLVR data by exploiting abundant, reasoning-rich, unverifiable internet text.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22975</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 13:39:11 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Mechanistic Data Attribution: Tracing the Training Origins of Interpretable LLM Units</title>
|
||||
<link>https://arxiv.org/abs/2601.21996</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.21996.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jianhui Chen, Yuzhang Luo, Liangming Pan</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> While Mechanistic Interpretability has identified interpretable circuits in LLMs, their causal origins in training data remain elusive. We introduce Mechanistic Data Attribution (MDA), a scalable framework that employs Influence Functions to trace interpretable units back to specific training samples. Through extensive experiments on the Pythia family, we causally validate that targeted intervention--removing or augmenting a small fraction of high-influence samples--significantly modulates the emergence of interpretable heads, whereas random interventions show no effect. Our analysis reveals that repetitive structural data (e.g., LaTeX, XML) acts as a mechanistic catalyst. Furthermore, we observe that interventions targeting induction head formation induce a concurrent change in the model's in-context learning (ICL) capability. This provides direct causal evidence for the long-standing hypothesis regarding the functional link between induction heads and ICL. Finally, we propose a mechanistic data augmentation pipeline that consistently accelerates circuit convergence across model scales, providing a principled methodology for steering the developmental trajectories of LLMs.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.21996</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 17:06:54 +0000</pubDate>
|
||||
<title>Machine Learning for Energy-Performance-aware Scheduling</title>
|
||||
<link>https://arxiv.org/abs/2601.23134</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23134.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zheyuan Hu, Yifei Shi</p><p><b>Upvotes:</b> 1</p><p><b>Summary:</b> In the post-Dennard era, optimizing embedded systems requires navigating complex trade-offs between energy efficiency and latency. Traditional heuristic tuning is often inefficient in such high-dimensional, non-smooth landscapes. In this work, we propose a Bayesian Optimization framework using Gaussian Processes to automate the search for optimal scheduling configurations on heterogeneous multi-core architectures. We explicitly address the multi-objective nature of the problem by approximating the Pareto Frontier between energy and time. Furthermore, by incorporating Sensitivity Analysis (fANOVA) and comparing different covariance kernels (e.g., Matérn vs. RBF), we provide physical interpretability to the black-box model, revealing the dominant hardware parameters driving system performance.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23134</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 16:23:06 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>PLANING: A Loosely Coupled Triangle-Gaussian Framework for Streaming 3D Reconstruction</title>
|
||||
<link>https://arxiv.org/abs/2601.22046</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22046.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Changjian Jiang, Kerui Ren, Xudong Li, Kaiwen Song, Linning Xu, Tao Lu, Junting Dong, Yu Zhang, Bo Dai, Mulin Yu</p><p><b>Upvotes:</b> 20</p><p><b>Summary:</b> Streaming reconstruction from monocular image sequences remains challenging, as existing methods typically favor either high-quality rendering or accurate geometry, but rarely both. We present PLANING, an efficient on-the-fly reconstruction framework built on a hybrid representation that loosely couples explicit geometric primitives with neural Gaussians, enabling geometry and appearance to be modeled in a decoupled manner. This decoupling supports an online initialization and optimization strategy that separates geometry and appearance updates, yielding stable streaming reconstruction with substantially reduced structural redundancy. PLANING improves dense mesh Chamfer-L2 by 18.52% over PGSR, surpasses ARTDECO by 1.31 dB PSNR, and reconstructs ScanNetV2 scenes in under 100 seconds, over 5x faster than 2D Gaussian Splatting, while matching the quality of offline per-scene optimization. Beyond reconstruction quality, the structural clarity and computational efficiency of \modelname~make it well suited for a broad range of downstream applications, such as enabling large-scale scene modeling and simulation-ready environments for embodied AI. Project page: https://city-super.github.io/PLANING/ .</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22046</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 17:47:26 +0000</pubDate>
|
||||
<title>THINKSAFE: Self-Generated Safety Alignment for Reasoning Models</title>
|
||||
<link>https://arxiv.org/abs/2601.23143</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23143.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Seanie Lee, Sangwoo Park, Yumin Choi, Gyeongman Kim, Minki Kang, Jihun Yun, Dongmin Park, Jongho Park, Sung Ju Hwang</p><p><b>Upvotes:</b> 32</p><p><b>Summary:</b> Large reasoning models (LRMs) achieve remarkable performance by leveraging reinforcement learning (RL) on reasoning tasks to generate long chain-of-thought (CoT) reasoning. However, this over-optimization often prioritizes compliance, making models vulnerable to harmful prompts. To mitigate this safety degradation, recent approaches rely on external teacher distillation, yet this introduces a distributional discrepancy that degrades native reasoning. We propose ThinkSafe, a self-generated alignment framework that restores safety alignment without external teachers. Our key insight is that while compliance suppresses safety mechanisms, models often retain latent knowledge to identify harm. ThinkSafe unlocks this via lightweight refusal steering, guiding the model to generate in-distribution safety reasoning traces. Fine-tuning on these self-generated responses effectively realigns the model while minimizing distribution shift. Experiments on DeepSeek-R1-Distill and Qwen3 show ThinkSafe significantly improves safety while preserving reasoning proficiency. Notably, it achieves superior safety and comparable reasoning to GRPO, with significantly reduced computational cost. Code, models, and datasets are available at https://github.com/seanie12/ThinkSafe.git.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23143</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 16:31:02 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>MetricAnything: Scaling Metric Depth Pretraining with Noisy Heterogeneous Sources</title>
|
||||
<link>https://arxiv.org/abs/2601.22054</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22054.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Baorui Ma, Jiahui Yang, Donglin Di, Xuancheng Zhang, Jianxun Cui, Hao Li, Yan Xie, Wei Chen</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Scaling has powered recent advances in vision foundation models, yet extending this paradigm to metric depth estimation remains challenging due to heterogeneous sensor noise, camera-dependent biases, and metric ambiguity in noisy cross-source 3D data. We introduce Metric Anything, a simple and scalable pretraining framework that learns metric depth from noisy, diverse 3D sources without manually engineered prompts, camera-specific modeling, or task-specific architectures. Central to our approach is the Sparse Metric Prompt, created by randomly masking depth maps, which serves as a universal interface that decouples spatial reasoning from sensor and camera biases. Using about 20M image-depth pairs spanning reconstructed, captured, and rendered 3D data across 10000 camera models, we demonstrate-for the first time-a clear scaling trend in the metric depth track. The pretrained model excels at prompt-driven tasks such as depth completion, super-resolution and Radar-camera fusion, while its distilled prompt-free student achieves state-of-the-art results on monocular depth estimation, camera intrinsics recovery, single/multi-view metric 3D reconstruction, and VLA planning. We also show that using pretrained ViT of Metric Anything as a visual encoder significantly boosts Multimodal Large Language Model capabilities in spatial intelligence. These results show that metric depth estimation can benefit from the same scaling laws that drive modern foundation models, establishing a new path toward scalable and efficient real-world metric perception. We open-source MetricAnything at http://metric-anything.github.io/metric-anything-io/ to support community research.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22054</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 17:52:41 +0000</pubDate>
|
||||
<title>DIFFA-2: A Practical Diffusion Large Language Model for General Audio Understanding</title>
|
||||
<link>https://arxiv.org/abs/2601.23161</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23161.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiaming Zhou, Xuxin Cheng, Shiwan Zhao, Yuhang Jia, Cao Liu, Ke Zeng, Xunliang Cai, Yong Qin</p><p><b>Upvotes:</b> 5</p><p><b>Summary:</b> Autoregressive (AR) large audio language models (LALMs) such as Qwen-2.5-Omni have achieved strong performance on audio understanding and interaction, but scaling them remains costly in data and computation, and strictly sequential decoding limits inference efficiency. Diffusion large language models (dLLMs) have recently been shown to make effective use of limited training data, and prior work on DIFFA indicates that replacing an AR backbone with a diffusion counterpart can substantially improve audio understanding under matched settings, albeit at a proof-of-concept scale without large-scale instruction tuning, preference alignment, or practical decoding schemes. We introduce DIFFA-2, a practical diffusion-based LALM for general audio understanding. DIFFA-2 upgrades the speech encoder, employs dual semantic and acoustic adapters, and is trained with a four-stage curriculum that combines semantic and acoustic alignment, large-scale supervised fine-tuning, and variance-reduced preference optimization, using only fully open-source corpora. Experiments on MMSU, MMAU, and MMAR show that DIFFA-2 consistently improves over DIFFA and is competitive to strong AR LALMs under practical training budgets, supporting diffusion-based modeling is a viable backbone for large-scale audio understanding. Our code is available at https://github.com/NKU-HLT/DIFFA.git.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23161</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 16:44:23 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>VTC-R1: Vision-Text Compression for Efficient Long-Context Reasoning</title>
|
||||
<link>https://arxiv.org/abs/2601.22069</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22069.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yibo Wang, Yongcheng Jing, Shunyu Liu, Hao Guan, Rong-cheng Tu, Chengyu Wang, Jun Huang, Dacheng Tao</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Long-context reasoning has significantly empowered large language models (LLMs) to tackle complex tasks, yet it introduces severe efficiency bottlenecks due to the computational complexity. Existing efficient approaches often rely on complex additional training or external models for compression, which limits scalability and discards critical fine-grained information. In this paper, we propose VTC-R1, a new efficient reasoning paradigm that integrates vision-text compression into the reasoning process. Instead of processing lengthy textual traces, VTC-R1 renders intermediate reasoning segments into compact images, which are iteratively fed back into vision-language models as "optical memory." We construct a training dataset based on OpenR1-Math-220K achieving 3.4x token compression and fine-tune representative VLMs-Glyph and Qwen3-VL. Extensive experiments on benchmarks such as MATH500, AIME25, AMC23 and GPQA-D demonstrate that VTC-R1 consistently outperforms standard long-context reasoning. Furthermore, our approach significantly improves inference efficiency, achieving 2.7x speedup in end-to-end latency, highlighting its potential as a scalable solution for reasoning-intensive applications. Our code is available at https://github.com/w-yibo/VTC-R1.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22069</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:07:39 +0000</pubDate>
|
||||
<title>FourierSampler: Unlocking Non-Autoregressive Potential in Diffusion Language Models via Frequency-Guided Generation</title>
|
||||
<link>https://arxiv.org/abs/2601.23182</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23182.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Siyang He, Qiqi Wang, Xiaoran Liu, Hongnan Ma, Yiwei Shi, Yuerong Song, Ying Zhu, Tianyi Liang, Zengfeng Huang, Ziwei He, Xipeng Qiu</p><p><b>Upvotes:</b> 18</p><p><b>Summary:</b> Despite the non-autoregressive potential of diffusion language models (dLLMs), existing decoding strategies demonstrate positional bias, failing to fully unlock the potential of arbitrary generation. In this work, we delve into the inherent spectral characteristics of dLLMs and present the first frequency-domain analysis showing that low-frequency components in hidden states primarily encode global structural information and long-range dependencies, while high-frequency components are responsible for characterizing local details. Based on this observation, we propose FourierSampler, which leverages a frequency-domain sliding window mechanism to dynamically guide the model to achieve a "structure-to-detail" generation. FourierSampler outperforms other inference enhancement strategies on LLADA and SDAR, achieving relative improvements of 20.4% on LLaDA1.5-8B and 16.0% on LLaDA-8B-Instruct. It notably surpasses similarly sized autoregressive models like Llama3.1-8B-Instruct.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23182</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 17:06:41 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Latent Adversarial Regularization for Offline Preference Optimization</title>
|
||||
<link>https://arxiv.org/abs/2601.22083</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22083.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Enyi Jiang, Yibo Jacky Zhang, Yinglun Xu, Andreas Haupt, Nancy Amato, Sanmi Koyejo</p><p><b>Upvotes:</b> 11</p><p><b>Summary:</b> Learning from human feedback typically relies on preference optimization that constrains policy updates through token-level regularization. However, preference optimization for language models is particularly challenging because token-space similarity does not imply semantic or behavioral similarity. To address this challenge, we leverage latent-space regularization for language model preference optimization. We introduce GANPO, which achieves latent-space regularization by penalizing divergence between the internal representations of a policy model and a reference model. Given that latent representations are not associated with explicit probability densities, we adopt an adversarial approach inspired by GANs to minimize latent-space divergence. We integrate GANPO as a regularizer into existing offline preference optimization objectives. Experiments across multiple model architectures and tasks show consistent improvements from latent-space regularization. Further, by comparing GANPO-induced inferential biases with those from token-level regularization, we find that GANPO provides more robust structural feedback under distributional shift and noise while maintaining comparable downstream performance with minor computational overhead.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22083</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:21:57 +0000</pubDate>
|
||||
<title>ReGuLaR: Variational Latent Reasoning Guided by Rendered Chain-of-Thought</title>
|
||||
<link>https://arxiv.org/abs/2601.23184</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23184.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Fanmeng Wang, Haotian Liu, Guojiang Zhao, Hongteng Xu, Zhifeng Gao</p><p><b>Upvotes:</b> 12</p><p><b>Summary:</b> While Chain-of-Thought (CoT) significantly enhances the performance of Large Language Models (LLMs), explicit reasoning chains introduce substantial computational redundancy. Recent latent reasoning methods attempt to mitigate this by compressing reasoning processes into latent space, but often suffer from severe performance degradation due to the lack of appropriate compression guidance. In this study, we propose Rendered CoT-Guided variational Latent Reasoning (ReGuLaR), a simple yet novel latent learning paradigm resolving this issue. Fundamentally, we formulate latent reasoning within the Variational Auto-Encoding (VAE) framework, sampling the current latent reasoning state from the posterior distribution conditioned on previous ones. Specifically, when learning this variational latent reasoning model, we render explicit reasoning chains as images, from which we extract dense visual-semantic representations to regularize the posterior distribution, thereby achieving efficient compression with minimal information loss. Extensive experiments demonstrate that ReGuLaR significantly outperforms existing latent reasoning methods across both computational efficiency and reasoning effectiveness, and even surpasses CoT through multi-modal reasoning, providing a new and insightful solution to latent reasoning. Code: https://github.com/FanmengWang/ReGuLaR.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23184</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 17:08:06 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>ECO: Quantized Training without Full-Precision Master Weights</title>
|
||||
<link>https://arxiv.org/abs/2601.22101</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22101.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mahdi Nikdan, Amir Zandieh, Dan Alistarh, Vahab Mirrokni</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Quantization has significantly improved the compute and memory efficiency of Large Language Model (LLM) training. However, existing approaches still rely on accumulating their updates in high-precision: concretely, gradient updates must be applied to a high-precision weight buffer, known as master weights. This buffer introduces substantial memory overhead, particularly for Sparse Mixture of Experts (SMoE) models, where model parameters and optimizer states dominate memory usage. To address this, we introduce the Error-Compensating Optimizer (ECO), which eliminates master weights by applying updates directly to quantized parameters. ECO quantizes weights after each step and carefully injects the resulting quantization error into the optimizer momentum, forming an error-feedback loop with no additional memory. We prove that, under standard assumptions and a decaying learning rate, ECO converges to a constant-radius neighborhood of the optimum, while naive master-weight removal can incur an error that is inversely proportional to the learning rate. We show empirical results for pretraining small Transformers (30-800M), a Gemma-3 1B model, and a 2.1B parameter Sparse MoE model with FP8 quantization, and fine-tuning DeepSeek-MoE-16B in INT4 precision. Throughout, ECO matches baselines with master weights up to near-lossless accuracy, significantly shifting the static memory vs validation loss Pareto frontier.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22101</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:35:01 +0000</pubDate>
|
||||
<title>Deep Search with Hierarchical Meta-Cognitive Monitoring Inspired by Cognitive Neuroscience</title>
|
||||
<link>https://arxiv.org/abs/2601.23188</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23188.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhongxiang Sun, Qipeng Wang, Weijie Yu, Jingxuan Yang, Haolang Lu, Jun Xu</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Deep search agents powered by large language models have demonstrated strong capabilities in multi-step retrieval, reasoning, and long-horizon task execution. However, their practical failures often stem from the lack of mechanisms to monitor and regulate reasoning and retrieval states as tasks evolve under uncertainty. Insights from cognitive neuroscience suggest that human metacognition is hierarchically organized, integrating fast anomaly detection with selectively triggered, experience-driven reflection. In this work, we propose Deep Search with Meta-Cognitive Monitoring (DS-MCM), a deep search framework augmented with an explicit hierarchical metacognitive monitoring mechanism. DS-MCM integrates a Fast Consistency Monitor, which performs lightweight checks on the alignment between external evidence and internal reasoning confidence, and a Slow Experience-Driven Monitor, which is selectively activated to guide corrective intervention based on experience memory from historical agent trajectories. By embedding monitoring directly into the reasoning-retrieval loop, DS-MCM determines both when intervention is warranted and how corrective actions should be informed by prior experience. Experiments across multiple deep search benchmarks and backbone models demonstrate that DS-MCM consistently improves performance and robustness.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23188</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 17:10:48 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>JUST-DUB-IT: Video Dubbing via Joint Audio-Visual Diffusion</title>
|
||||
<link>https://arxiv.org/abs/2601.22143</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22143.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Anthony Chen, Naomi Ken Korem, Tavi Halperin, Matan Ben Yosef, Urska Jelercic, Ofir Bibi, Or Patashnik, Daniel Cohen-Or</p><p><b>Upvotes:</b> 3</p><p><b>Summary:</b> Audio-Visual Foundation Models, which are pretrained to jointly generate sound and visual content, have recently shown an unprecedented ability to model multi-modal generation and editing, opening new opportunities for downstream tasks. Among these tasks, video dubbing could greatly benefit from such priors, yet most existing solutions still rely on complex, task-specific pipelines that struggle in real-world settings. In this work, we introduce a single-model approach that adapts a foundational audio-video diffusion model for video-to-video dubbing via a lightweight LoRA. The LoRA enables the model to condition on an input audio-video while jointly generating translated audio and synchronized facial motion. To train this LoRA, we leverage the generative model itself to synthesize paired multilingual videos of the same speaker. Specifically, we generate multilingual videos with language switches within a single clip, and then inpaint the face and audio in each half to match the language of the other half. By leveraging the rich generative prior of the audio-visual model, our approach preserves speaker identity and lip synchronization while remaining robust to complex motion and real-world dynamics. We demonstrate that our approach produces high-quality dubbed videos with improved visual fidelity, lip synchronization, and robustness compared to existing dubbing pipelines.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22143</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:57:13 +0000</pubDate>
|
||||
<title>Scaling Multiagent Systems with Process Rewards</title>
|
||||
<link>https://arxiv.org/abs/2601.23228</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23228.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ed Li, Junyu Ren, Cat Yan</p><p><b>Upvotes:</b> 5</p><p><b>Summary:</b> While multiagent systems have shown promise for tackling complex tasks via specialization, finetuning multiple agents simultaneously faces two key challenges: (1) credit assignment across agents, and (2) sample efficiency of expensive multiagent rollouts. In this work, we propose finetuning multiagent systems with per-action process rewards from AI feedback (MAPPA) to address both. Through assigning credit to individual agent actions rather than only at task completion, MAPPA enables fine-grained supervision without ground truth labels while extracting maximal training signal from each rollout. We demonstrate our approach on competition math problems and tool-augmented data analysis tasks. On unseen math problems, MAPPA achieves +5.0--17.5pp on AIME and +7.8--17.2pp on AMC. For data analysis tasks, our method improves success rate by +12.5pp while quality metrics improve by up to 30%, validating that per-action supervision can lead to improvements across different multiagent system on various domains. By addressing these challenges, our work takes a first step toward scaling multiagent systems for complex, long-horizon tasks with minimal human supervision.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23228</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 17:55:27 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>FineInstructions: Scaling Synthetic Instructions to Pre-Training Scale</title>
|
||||
<link>https://arxiv.org/abs/2601.22146</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22146.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ajay Patel, Colin Raffel, Chris Callison-Burch</p><p><b>Upvotes:</b> 4</p><p><b>Summary:</b> Due to limited supervised training data, large language models (LLMs) are typically pre-trained via a self-supervised "predict the next word" objective on a vast amount of unstructured text data. To make the resulting model useful to users, it is further trained on a far smaller amount of "instruction-tuning" data comprised of supervised training examples of instructions and responses. To overcome the limited amount of supervised data, we propose a procedure that can transform the knowledge in internet-scale pre-training documents into billions of synthetic instruction and answer training pairs. The resulting dataset, called FineInstructions, uses ~18M instruction templates created from real user-written queries and prompts. These instruction templates are matched to and instantiated with human-written source documents from unstructured pre-training corpora. With "supervised" synthetic training data generated at this scale, an LLM can be pre-trained from scratch solely with the instruction-tuning objective, which is far more in-distribution with the expected downstream usage of LLMs (responding to user prompts). We conduct controlled token-for-token training experiments and find pre-training on FineInstructions outperforms standard pre-training and other proposed synthetic pre-training techniques on standard benchmarks measuring free-form response quality. Our resources can be found at https://huggingface.co/fineinstructions .</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22146</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:58:47 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>DynamicVLA: A Vision-Language-Action Model for Dynamic Object Manipulation</title>
|
||||
<link>https://arxiv.org/abs/2601.22153</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22153.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haozhe Xie, Beichen Wen, Jiarui Zheng, Zhaoxi Chen, Fangzhou Hong, Haiwen Diao, Ziwei Liu</p><p><b>Upvotes:</b> 60</p><p><b>Summary:</b> Manipulating dynamic objects remains an open challenge for Vision-Language-Action (VLA) models, which, despite strong generalization in static manipulation, struggle in dynamic scenarios requiring rapid perception, temporal anticipation, and continuous control. We present DynamicVLA, a framework for dynamic object manipulation that integrates temporal reasoning and closed-loop adaptation through three key designs: 1) a compact 0.4B VLA using a convolutional vision encoder for spatially efficient, structurally faithful encoding, enabling fast multimodal inference; 2) Continuous Inference, enabling overlapping reasoning and execution for lower latency and timely adaptation to object motion; and 3) Latent-aware Action Streaming, which bridges the perception-execution gap by enforcing temporally aligned action execution. To fill the missing foundation of dynamic manipulation data, we introduce the Dynamic Object Manipulation (DOM) benchmark, built from scratch with an auto data collection pipeline that efficiently gathers 200K synthetic episodes across 2.8K scenes and 206 objects, and enables fast collection of 2K real-world episodes without teleoperation. Extensive evaluations demonstrate remarkable improvements in response speed, perception, and generalization, positioning DynamicVLA as a unified framework for general dynamic object manipulation across embodiments.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22153</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:59:51 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Exploring Reasoning Reward Model for Agents</title>
|
||||
<link>https://arxiv.org/abs/2601.22154</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22154.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Kaixuan Fan, Kaituo Feng, Manyuan Zhang, Tianshuo Peng, Zhixun Li, Yilei Jiang, Shuang Chen, Peng Pei, Xunliang Cai, Xiangyu Yue</p><p><b>Upvotes:</b> 20</p><p><b>Summary:</b> Agentic Reinforcement Learning (Agentic RL) has achieved notable success in enabling agents to perform complex reasoning and tool use. However, most methods still relies on sparse outcome-based reward for training. Such feedback fails to differentiate intermediate reasoning quality, leading to suboptimal training results. In this paper, we introduce Agent Reasoning Reward Model (Agent-RRM), a multi-faceted reward model that produces structured feedback for agentic trajectories, including (1) an explicit reasoning trace , (2) a focused critique that provides refinement guidance by highlighting reasoning flaws, and (3) an overall score that evaluates process performance. Leveraging these signals, we systematically investigate three integration strategies: Reagent-C (text-augmented refinement), Reagent-R (reward-augmented guidance), and Reagent-U (unified feedback integration). Extensive evaluations across 12 diverse benchmarks demonstrate that Reagent-U yields substantial performance leaps, achieving 43.7% on GAIA and 46.2% on WebWalkerQA, validating the effectiveness of our reasoning reward model and training schemes. Code, models, and datasets are all released to facilitate future research.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22154</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:59:52 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Hybrid Linear Attention Done Right: Efficient Distillation and Effective Architectures for Extremely Long Contexts</title>
|
||||
<link>https://arxiv.org/abs/2601.22156</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22156.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yingfa Chen, Zhen Leng Thai, Zihan Zhou, Zhu Zhang, Xingyu Shen, Shuo Wang, Chaojun Xiao, Xu Han, Zhiyuan Liu</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Hybrid Transformer architectures, which combine softmax attention blocks and recurrent neural networks (RNNs), have shown a desirable performance-throughput tradeoff for long-context modeling, but their adoption and studies are hindered by the prohibitive cost of large-scale pre-training from scratch. Some recent studies have shown that pre-trained softmax attention blocks can be converted into RNN blocks through parameter transfer and knowledge distillation. However, these transfer methods require substantial amounts of training data (more than 10B tokens), and the resulting hybrid models also exhibit poor long-context performance, which is the scenario where hybrid models enjoy significant inference speedups over Transformer-based models. In this paper, we present HALO (Hybrid Attention via Layer Optimization), a pipeline for distilling Transformer models into RNN-attention hybrid models. We then present HypeNet, a hybrid architecture with superior length generalization enabled by a novel position encoding scheme (named HyPE) and various architectural modifications. We convert the Qwen3 series into HypeNet using HALO, achieving performance comparable to the original Transformer models while enjoying superior long-context performance and efficiency. The conversion requires just 2.3B tokens, less than 0.01% of their pre-training data</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22156</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:59:53 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>Discovering Hidden Gems in Model Repositories</title>
|
||||
<link>https://arxiv.org/abs/2601.22157</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22157.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jonathan Kahana, Eliahu Horwitz, Yedid Hoshen</p><p><b>Upvotes:</b> 16</p><p><b>Summary:</b> Public repositories host millions of fine-tuned models, yet community usage remains disproportionately concentrated on a small number of foundation checkpoints. We investigate whether this concentration reflects efficient market selection or if superior models are systematically overlooked. Through an extensive evaluation of over 2,000 models, we show the prevalence of "hidden gems", unpopular fine-tunes that significantly outperform their popular counterparts. Notably, within the Llama-3.1-8B family, we find rarely downloaded checkpoints that improve math performance from 83.2% to 96.0% without increasing inference costs. However, discovering these models through exhaustive evaluation of every uploaded model is computationally infeasible. We therefore formulate model discovery as a Multi-Armed Bandit problem and accelerate the Sequential Halving search algorithm by using shared query sets and aggressive elimination schedules. Our method retrieves top models with as few as 50 queries per candidate, accelerating discovery by over 50x.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22157</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:59:55 +0000</pubDate>
|
||||
</item>
|
||||
<item>
|
||||
<title>One-step Latent-free Image Generation with Pixel Mean Flows</title>
|
||||
<link>https://arxiv.org/abs/2601.22158</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.22158.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yiyang Lu, Susie Lu, Qiao Sun, Hanhong Zhao, Zhicheng Jiang, Xianbang Wang, Tianhong Li, Zhengyang Geng, Kaiming He</p><p><b>Upvotes:</b> 7</p><p><b>Summary:</b> Modern diffusion/flow-based models for image generation typically exhibit two core characteristics: (i) using multi-step sampling, and (ii) operating in a latent space. Recent advances have made encouraging progress on each aspect individually, paving the way toward one-step diffusion/flow without latents. In this work, we take a further step towards this goal and propose "pixel MeanFlow" (pMF). Our core guideline is to formulate the network output space and the loss space separately. The network target is designed to be on a presumed low-dimensional image manifold (i.e., x-prediction), while the loss is defined via MeanFlow in the velocity space. We introduce a simple transformation between the image manifold and the average velocity field. In experiments, pMF achieves strong results for one-step latent-free generation on ImageNet at 256x256 resolution (2.22 FID) and 512x512 resolution (2.48 FID), filling a key missing piece in this regime. We hope that our study will further advance the boundaries of diffusion/flow-based generative models.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.22158</guid>
|
||||
<pubDate>Thu, 29 Jan 2026 18:59:56 +0000</pubDate>
|
||||
<title>PaperBanana: Automating Academic Illustration for AI Scientists</title>
|
||||
<link>https://arxiv.org/abs/2601.23265</link>
|
||||
<description><p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2601.23265.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Dawei Zhu, Rui Meng, Yale Song, Xiyu Wei, Sujian Li, Tomas Pfister, Jinsung Yoon</p><p><b>Upvotes:</b> 18</p><p><b>Summary:</b> Despite rapid advances in autonomous AI scientists powered by language models, generating publication-ready illustrations remains a labor-intensive bottleneck in the research workflow. To lift this burden, we introduce PaperBanana, an agentic framework for automated generation of publication-ready academic illustrations. Powered by state-of-the-art VLMs and image generation models, PaperBanana orchestrates specialized agents to retrieve references, plan content and style, render images, and iteratively refine via self-critique. To rigorously evaluate our framework, we introduce PaperBananaBench, comprising 292 test cases for methodology diagrams curated from NeurIPS 2025 publications, covering diverse research domains and illustration styles. Comprehensive experiments demonstrate that PaperBanana consistently outperforms leading baselines in faithfulness, conciseness, readability, and aesthetics. We further show that our method effectively extends to the generation of high-quality statistical plots. Collectively, PaperBanana paves the way for the automated generation of publication-ready illustrations.</p></description>
|
||||
<guid isPermaLink="false">https://arxiv.org/abs/2601.23265</guid>
|
||||
<pubDate>Fri, 30 Jan 2026 18:33:37 +0000</pubDate>
|
||||
</item>
|
||||
</channel>
|
||||
</rss>
|
||||
|
||||
Reference in New Issue
Block a user