❌

Normal view

Received — 27 May 2026 ⏭
  • ✇cs.AI, q-bio.NC updates on arXiv.org
  • Extreme Region Policy Distillation Changyu Chen · Xiting Wang · Rui Yan
    arXiv:2605.25582v1 Announce Type: cross Abstract: Reinforcement learning for large language models faces a fundamental trade-off between sample efficiency and asymptotic performance: strictly on-policy methods discard trajectories after a single update, while off-policy reuse introduces distribution mismatch that existing trust-region techniques mitigate primarily by enforcing conservative optimization, often leaving rich training signals underutilized. To investigate this, we perform extensive
     
  • 🏷️
  • ↗️

Extreme Region Policy Distillation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Changyu Chen · Xiting Wang · Rui Yan
26 May 2026 at 12:00
arXiv:2605.25582v1 Announce Type: cross Abstract: Reinforcement learning for large language models faces a fundamental trade-off between sample efficiency and asymptotic performance: strictly on-policy methods discard trajectories after a single update, while off-policy reuse introduces distribution mismatch that existing trust-region techniques mitigate primarily by enforcing conservative optimization, often leaving rich training signals underutilized. To investigate this, we perform extensive off-policy updates on fixed data. Our experiments reveal that aggressive multi-step optimization brings rapid initial gains, but excessive updates cause trajectory probabilities to deviate and entropy to collapse, with performance plateauing early. Tightening KL constraints merely lowers the ceiling without resolving the degradation. This motivates Extreme Region Policy Distillation (ERPD), a two-stage framework that decouples sample efficiency from KL efficiency. The first stage performs weakly constrained off-policy optimization on fixed data to maximally extract training signals. The resulting policy provides token-level supervision. In the second stage, we distill these signals into the base policy under trust-region constraints, filtering harmful drift while preserving useful signals. The distilled policy achieves comparable or better performance with substantially smaller KL divergence, indicating that much of the first-stage divergence was spent on unnecessary drift rather than genuine improvement. Crucially, ERPD accommodates both strong and weak teachers: when aggressive optimization yields no stronger policy, even degenerate teachers provide effective supervision via alternative signal construction strategies. We validate ERPD on mathematical reasoning, showing gains for strong base models where on-policy training plateaus, and reliable improvements with weak teachers.

MoBiQuant: Mixture-of-Bits Quantization for Token-Adaptive Any-Precision LLM

✇cs.AI, q-bio.NC updates on arXiv.org
By: Dongwei Wang · Jinhee Kim · Seokho Han · Denis Gudovskiy · Yohei Nakata · Tomoyuki Okuno · KhayTze Peong · Kang Eun Jeon · Jong Hwan Ko · Yiran Chen · Huanrui Yang
26 May 2026 at 12:00
arXiv:2602.20191v2 Announce Type: replace-cross Abstract: Dynamic runtime latency and memory constraints necessitate flexible large language model (LLM) deployment, where an LLM can be inferred with various quantization precisions based on available computational resources. Recent work on such any-precision quantization either relies on hardware-inefficient vector quantization or induces additional scaling factors when switching between bit-widths. Meanwhile, existing post-training quantization (PTQ) methods calibrated for a fixed low precision show poor generalizability under runtime precision change. In this work, we attribute the source of poor generalization across bit-widths to a precision-dependent \textit{outlier migration} phenomenon where the distribution of PTQ-sensitive tokens changes across precisions. Motivated by this observation, we propose \texttt{MoBiQuant}, a novel any-precision Mixture-of-Bits quantization framework that adjusts weight precision for flexible LLM inference based on token sensitivity. Specifically, we propose a many-in-one recursive residual quantization that can iteratively reconstruct higher-precision weights at runtime and mitigates \textit{outlier migration} with a token-aware router to dynamically select the optimal inference precision of each token.Extensive experiments show that \texttt{MoBiQuant} matches or surpasses frontier single-precision PTQ while exhibiting strong elasticity, achieving significant memory savings and throughput gains of up to $1.34\times$ over state-of-the-art any-precision methods.

A SAUR gene enhances maize drought resilience by promoting silk elongation

✇Nature - Issue - nature.com science feeds
By: Chaohui Zhu · Zhirui Yang · Shiping Yang · Xueyan Zhou · Boxin Liu · Tian Tian · Bochen Zhao · Yingying Xie · Yujun Liu · Jinkui Cheng · Huaijun Tang · Yanjun Zhang · Xiaoqing Xie · Lei Zhang · Cheng Liu · Xingrong Wang · Shengxue Liu · Feng Qin
20 May 2026 at 08:00

Nature, Published online: 20 May 2026; doi:10.1038/s41586-026-10566-9

The Small Auxin Up RNA (SAUR) protein ZmSAUR72 in maize (Zea mays) promotes silk growth via regulation of H+-ATPase activity, and is a key determinant of the anthesis-silking interval and thus resilience to drought.
Received — 2 April 2026 ⏭

SemLoc: Structured Grounding of Free-Form LLM Reasoning for Fault Localization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Zhaorui Yang · Haichao Zhu · Qian Zhang · Rajiv Gupta · Ashish Kundu
1 April 2026 at 12:00
arXiv:2603.29109v1 Announce Type: cross Abstract: Fault localization identifies program locations responsible for observed failures. Existing techniques rank suspicious code using syntactic spectra--signals derived from execution structure such as statement coverage, control-flow divergence, or dependency reachability. These signals collapse for semantic bugs, where failing and passing executions follow identical code paths and differ only in whether semantic intent is satisfied. Recent LLM-based approaches introduce semantic reasoning but produce stochastic, unverifiable outputs that cannot be systematically cross-referenced across tests or distinguish root causes from cascading effects. We present SemLoc, a fault localization framework based on structured semantic grounding. SemLoc converts free-form LLM reasoning into a closed intermediate representation that binds each inferred property to a typed program anchor, enabling runtime checking and attribution to program structure. It executes instrumented programs to construct a semantic violation spectrum--a constraint-by-test matrix--from which suspiciousness scores are derived analogously to coverage-based methods. A counterfactual verification step further prunes over-approximate constraints and isolates primary causal violations. We evaluate SemLoc on SemFault-250, a corpus of 250 Python programs with single semantic faults. SemLoc outperforms five coverage-, reduction-, and LLM-based baselines, achieving Top-1 accuracy of 42.8% and Top-3 of 68%, while reducing inspection to 7.6% of executable lines. Counterfactual verification provides an additional 12% accuracy gain and identifies primary causal semantic constraints.
Received — 26 March 2026 ⏭

Unannotated noncoding transcripts as a source of intratumor heterogeneity in malignant cell states

✇(Multiomics OR Omics) AND (Lung OR gastric OR Hepatocellular)
By: Xiaolan Zhou · Limin Lin · Yu Chen · Guangrui Yang · Ting Ni · Zhaozhao Zhao
23 March 2026 at 18:00

Sci China Life Sci. 2026 Mar 16. doi: 10.1007/s11427-025-3273-6. Online ahead of print.

ABSTRACT

Phenotypic diversity of malignant cells within a tumor underlies intratumor heterogeneity (ITH), a key determinant of cancer metastasis and treatment failure. However, the molecular mechanisms driving this heterogeneity are poorly understood. Here, we curated and analyzed a cohort of 3' tag-based single-cell RNA-seq covering 12 common cancer types. We identified thousands of poly(A) site (PAS) peaks representing the 3' ends of previously unannotated transcripts, whose expression is widely associated with diverse malignant cellular states. By integrating multi-omics data, we characterized the expression patterns and epigenetic landscape of these unannotated PAS peak-associated transcripts (UPTs). The expression heterogeneity of UPTs was supported by multi-region sampling bulk RNA-seq data and recapitulated within cancer cell lines. As proof of principle validation, functional experiments confirmed that two noncoding UPTs promoted the proliferation and migration of lung cancer cells. Our results suggest that epigenetic activation of unannotated noncoding transcripts might represent a previously unrecognized mechanism contributing to transcriptomic ITH.

PMID:41870780 | DOI:10.1007/s11427-025-3273-6

Unannotated noncoding transcripts as a source of intratumor heterogeneity in malignant cell states

✇Omics In Lung
By: Xiaolan Zhou · Limin Lin · Yu Chen · Guangrui Yang · Ting Ni · Zhaozhao Zhao
23 March 2026 at 18:00

Sci China Life Sci. 2026 Mar 16. doi: 10.1007/s11427-025-3273-6. Online ahead of print.

ABSTRACT

Phenotypic diversity of malignant cells within a tumor underlies intratumor heterogeneity (ITH), a key determinant of cancer metastasis and treatment failure. However, the molecular mechanisms driving this heterogeneity are poorly understood. Here, we curated and analyzed a cohort of 3' tag-based single-cell RNA-seq covering 12 common cancer types. We identified thousands of poly(A) site (PAS) peaks representing the 3' ends of previously unannotated transcripts, whose expression is widely associated with diverse malignant cellular states. By integrating multi-omics data, we characterized the expression patterns and epigenetic landscape of these unannotated PAS peak-associated transcripts (UPTs). The expression heterogeneity of UPTs was supported by multi-region sampling bulk RNA-seq data and recapitulated within cancer cell lines. As proof of principle validation, functional experiments confirmed that two noncoding UPTs promoted the proliferation and migration of lung cancer cells. Our results suggest that epigenetic activation of unannotated noncoding transcripts might represent a previously unrecognized mechanism contributing to transcriptomic ITH.

PMID:41870780 | DOI:10.1007/s11427-025-3273-6

Received — 17 March 2026 ⏭

FAPE-IR: Frequency-Aware Planning and Execution Framework for All-in-One Image Restoration

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jingren Liu · Shuning Xu · Qirui Yang · Yun Wang · Xiangyu Chen · Zhong Ji
16 March 2026 at 12:00
arXiv:2511.14099v3 Announce Type: replace-cross Abstract: All-in-One Image Restoration (AIO-IR) aims to develop a unified model that can handle multiple degradations under complex conditions. However, existing methods often rely on task-specific designs or latent routing strategies, making it hard to adapt to real-world scenarios with various degradations. We propose FAPE-IR, a Frequency-Aware Planning and Execution framework for image restoration. It uses a frozen Multimodal Large Language Model (MLLM) as a planner to analyze degraded images and generate concise, frequency-aware restoration plans. These plans guide a LoRA-based Mixture-of-Experts (LoRA-MoE) module within a diffusion-based executor, which dynamically selects high- or low-frequency experts, complemented by frequency features of the input image. To further improve restoration quality and reduce artifacts, we introduce adversarial training and a frequency regularization loss. By coupling semantic planning with frequency-based restoration, FAPE-IR offers a unified and interpretable solution for all-in-one image restoration. Extensive experiments show that FAPE-IR achieves state-of-the-art performance across seven restoration tasks and exhibits strong zero-shot generalization under mixed degradations.
Received — 14 March 2026 ⏭

CDRRM: Contrast-Driven Rubric Generation for Reliable and Interpretable Reward Modeling

✇cs.AI, q-bio.NC updates on arXiv.org
By: Dengcan Liu · Fengkai Yang · Xiaohan Wang · Shurui Yan · Jiajun Chai · Jiahao Li · Yikun Ban · Zhendong Mao · Wei Lin · Guojun Yin
10 March 2026 at 12:00
arXiv:2603.08035v1 Announce Type: new Abstract: Reward modeling is essential for aligning Large Language Models(LLMs) with human preferences, yet conventional reward models suffer from poor interpretability and heavy reliance on costly expert annotations. While recent rubric-based approaches enhance evaluation transparency, they lack systematic quality control, yielding noisy and redundant criteria, failing to mitigate persistent biases (e.g., verbosity, position) in LLM evaluators, and creating a scalability-reliability trade-off. To address these limitations, we propose CDRRM (Contrast-Driven Rubric Reward Model), a framework built on a novel Contrast-then-Synthesis paradigm for high-quality rubric generation and guided preference judgment. CDRRM first conducts multi-dimensional contrastive profiling on preference pairs to identify causal discriminative factors, then synthesizes these insights into compact, context-aware rubrics to guide preference judg- ments. Extensive experiments on three authoritative benchmarks (RewardBench, RMBench, RMB) demonstrate that CDRRM achieves state-of-the-art performance across diverse domains and effectively mitigates aforementioned evaluation biases. Notably, our approach delivers exceptional data efficiency: training the rubric generator on only 3k high-quality samples empowers a frozen pre-trained judge model to outperform fully fine-tuned baselines. This work offers a scalable, interpretable, and data-efficient path for reward modeling.
Received — 5 March 2026 ⏭

Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy

✇cs.AI, q-bio.NC updates on arXiv.org
By: Chris Yuhao Liu · Liang Zeng · Yuzhen Xiao · Jujie He · Jiacai Liu · Chaojie Wang · Rui Yan · Wei Shen · Fuxiang Zhang · Jiacheng Xu · Yang Liu · Yahui Zhou
4 March 2026 at 13:00
arXiv:2507.01352v3 Announce Type: replace-cross Abstract: Despite the critical role of reward models (RMs) in Reinforcement Learning from Human Feedback (RLHF), current state-of-the-art open RMs perform poorly on most existing evaluation benchmarks, failing to capture nuanced human preferences. We hypothesize that this brittleness stems primarily from limitations in preference datasets, which are often narrowly scoped, synthetically labeled, or lack rigorous quality control. To address these challenges, we present SynPref-40M, a large-scale preference dataset comprising 40 million preference pairs. To enable data curation at scale, we design a human-AI synergistic two-stage pipeline that leverages the complementary strengths of human annotation quality and AI scalability. In this pipeline, humans provide verified annotations, while LLMs perform automatic curation based on human guidance. Training on this preference mixture, we introduce Skywork-Reward-V2, a suite of eight reward models ranging from 0.6B to 8B parameters, trained on a carefully curated subset of 26 million preference pairs from SynPref-40M. We demonstrate that Skywork-Reward-V2 is versatile across a wide range of capabilities, including alignment with human preferences, objective correctness, safety, resistance to stylistic biases, and best-of-N scaling. These reward models achieve state-of-the-art performance across seven major reward model benchmarks, outperform generative reward models, and demonstrate strong downstream performance. Ablation studies confirm that effectiveness stems not only from data scale but also from high-quality curation. The Skywork-Reward-V2 series represents substantial progress in open reward models, demonstrating how human-AI curation synergy can unlock significantly higher data quality.
Received — 25 February 2026 ⏭

BEAT: Visual Backdoor Attacks on VLM-based Embodied Agents via Contrastive Trigger Learning

✇cs.AI, q-bio.NC updates on arXiv.org
By: Qiusi Zhan · Hyeonjeong Ha · Rui Yang · Sirui Xu · Hanyang Chen · Liang-Yan Gui · Yu-Xiong Wang · Huan Zhang · Heng Ji · Daniel Kang
24 February 2026 at 13:00
arXiv:2510.27623v3 Announce Type: replace Abstract: Recent advances in Vision-Language Models (VLMs) have propelled embodied agents by enabling direct perception, reasoning, and planning task-oriented actions from visual inputs. However, such vision-driven embodied agents open a new attack surface: visual backdoor attacks, where the agent behaves normally until a visual trigger appears in the scene, then persistently executes an attacker-specified multi-step policy. We introduce BEAT, the first framework to inject such visual backdoors into VLM-based embodied agents using objects in the environments as triggers. Unlike textual triggers, object triggers exhibit wide variation across viewpoints and lighting, making them difficult to implant reliably. BEAT addresses this challenge by (1) constructing a training set that spans diverse scenes, tasks, and trigger placements to expose agents to trigger variability, and (2) introducing a two-stage training scheme that first applies supervised fine-tuning (SFT) and then our novel Contrastive Trigger Learning (CTL). CTL formulates trigger discrimination as preference learning between trigger-present and trigger-free inputs, explicitly sharpening the decision boundaries to ensure precise backdoor activation. Across various embodied agent benchmarks and VLMs, BEAT achieves attack success rates up to 80%, while maintaining strong benign task performance, and generalizes reliably to out-of-distribution trigger placements. Notably, compared to naive SFT, CTL boosts backdoor activation accuracy up to 39% under limited backdoor data. These findings expose a critical yet unexplored security risk in VLM-based embodied agents, underscoring the need for robust defenses before real-world deployment.
There are no more articles
❌