❌

Normal view

Received — 10 September 2026 ⏭

Sequence and structural determinants of efficacious de novo chimaeric antigen receptors

✇Nature Biomedical Engineering
By: Arthur Chow · Hoyin Chu · Ruofan Li · Benan N. Nalbant · Abdul Vehab Dozic · Laura C. Kida · Zeyu Tang · Joseph R. Palmeri · Caleb A. Lareau
9 September 2026 at 08:00

Nature Biomedical Engineering, Published online: 09 September 2026; doi:10.1038/s41551-026-01790-9

A generative protein design workflow addresses important challenges with de novo protein engineering of chimaeric antigen receptors (CARs) to improve targeting of proteins important in cancer and create more effective CAR T therapies.

ViSR-KGC: Visual Subgraph Reasoning with Vision-Language Models for Multimodal Knowledge Graph Completion

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jiafan Li · Mengxue Yang · Jiaqi Zhu · Liang Chang · Ying Li · Hongan Wang
10 September 2026 at 12:00
arXiv:2608.05833v3 Announce Type: replace Abstract: Knowledge graph completion (KGC) aims to infer missing entities or relations from incomplete graph structures, and has evolved into multimodal knowledge graph completion (MMKGC), where entities are associated with multiple modalities such as text and images. Traditional representation learning approaches follow the embedding-based paradigm and may struggle when relation-specific evidence is limited. Meanwhile, LLM-based reasoning methods typically linearize graph structures into textual prompts, which obscures structural topology and neglects vital visual information. While vision-language models (VLMs) excel at multimodal reasoning, they cannot natively interpret structured graph topology, particularly when it comes to knowledge graphs where nodes and edges carry complex semantics. To bridge this gap, we propose ViSR-KGC, a visual subgraph reasoning approach for KGC. It integrates three complementary capabilities to capture semantic correlations: identifying global topology dependencies via representation learning, analyzing local multimodal evidence using VLMs, and providing necessary commonsense knowledge inherent in pre-trained models. Based on learned multimodal embeddings, our framework first extracts a compact and query-aware subgraph from the MMKG. Then, this subgraph is transformed into a visually interpretable image using a layout strategy selected through empirical comparison. Finally, the visualized subgraph, entity images, textual descriptions, and candidate answers are combined into a unified prompt, enabling the VLM to infer the missing entity.

Complete biosynthesis of the anticancer cephalotaxinone and homoerythratine

✇Cell
By: Runze Tian · Feifan Lin · Nianxin Guo · Chendi Liu · Kaiqi Chen · Yunxi Han · Ruiyun Lan · Qiang Li · Jianbin Yan · Xiaoguang Lei
11 June 2026 at 08:00
Complete biosynthetic pathways for cephalotaxinone and homoerythratine were elucidated from the endangered plant Cephalotaxus fortunei. Thirteen key enzymes were identified, including two homologous cytochrome P450 enzymes that catalyze a rare divergent oxidation process governing alkaloid scaffold diversification. Full pathway reconstitution in Nicotiana benthamiana establishes a foundation for the sustainable production of the anticancer agent homoharringtonine.
Received — 27 May 2026 ⏭

VectorArk: Learning Practical Image Vectorization with Rounded Polygon Representation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Tarun Gehlaut · Difan Liu · Charu Bansal · Krutik Malani · Souymodip Chakraborty · Ankit Phogat · Matthew Fisher · Vineet Batra
26 May 2026 at 12:00
arXiv:2605.24398v1 Announce Type: cross Abstract: Recent vision-language model (VLM)-based approaches have achieved impressive results on image vectorization tasks. However, they are typically evaluated on synthetic benchmarks, where clean SVGs are rasterized at high resolution and then re-vectorized. As a result, these methods generalize poorly to real-world scenarios, such as images with unknown rasterization methods or those generated by text-to-image models. We introduce VectorArk, a new VLM-based model designed for robust and practical image vectorization. VectorArk employs a novel rounded polygon representation that simplifies the learning process while naturally producing smooth, visually appealing primitives. We also propose a degradation model that enhances robustness across diverse and imperfect inputs. Our experiments show that, in contrast to previous methods, VectorArk achieves superior geometric completeness and artifact suppression across multiple datasets, with comprehensive ablations validating the contribution of each component.

Step-TP: A Grounded, Step-Level Dataset with Chain-of-Thought Reasoning for LLM-Guided Tensor Program Optimization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Mengfan Liu · Da Zheng · Junwei Su · Chuan Wu
26 May 2026 at 12:00
arXiv:2605.25954v1 Announce Type: cross Abstract: Despite the strong reasoning capabilities of large language models (LLMs), optimizing the execution efficiency of tensor programs remains challenging due to the need for precise, composable transformation decisions. Recent LLM-guided approaches frame tensor program optimization as an iterative decision process, but existing datasets provide only end-to-end optimized program pairs using token-inefficient representations, lacking verifiable step-level supervision and interpretability. As a result, LLMs struggle to make reliable single-step decisions in large combinatorial optimization spaces. We introduce Step-TP, a post-training dataset for tensor program optimization that provides grounded, atomic, step-level supervision with structured chain-of-thought (CoT) reasoning. Step-TP forms a closed reasoning loop over intermediate program states, enabling reliable multi-step optimization rather than outcome imitation. Its design is guided by four principles: (i) a token-efficient, verifiable intermediate representation (IR) that deterministically lowers to TVM TIR; (ii) atomic and composable optimization strategies that decompose complex trajectories into interpretable single-step decisions; (iii) structured CoT supervision coupled with explicit IR-to-IR state transitions; and (iv) strategy filtering to balance coverage while preventing shortcut exploitation. The dataset and implementation are available at a GitHub link, https://github.com/LIUMENGFAN-gif/StepTP.

HiTeC: Hierarchical Contrastive Learning on Text-Attributed Hypergraph with Semantic-Aware Augmentation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Mengting Pan · Fan Li · Chen Chen · Xiaoyang Wang · Wenjie Zhang
26 May 2026 at 12:00
arXiv:2508.03104v3 Announce Type: replace-cross Abstract: Contrastive learning (CL) has become a dominant paradigm for self-supervised hypergraph learning, enabling effective training without costly labels. However, node entities in real-world hypergraphs are often associated with rich textual information, which has been largely ignored in prior works. Directly applying existing CL-based methods to such text-attributed hypergraphs (TAHGs) leads to three key limitations: (1) The common use of graph-agnostic text encoders fails to capture the correlations between textual semantics and hypergraph topology, resulting in less expressive representations. (2) Their reliance on random data augmentations introduces noise and weakens the contrastive signals. (3) The primary focus on node- and hyperedge-level contrastive signals limits the ability to capture long-range dependencies, which is essential for effective representation learning. To address these challenges, we introduce HiTeC, a two-stage hierarchical contrastive learning framework for effective self-supervised learning on TAHGs. In the first stage, we pre-train the text encoder with a structure-aware contrastive objective to overcome the graph-agnostic nature of conventional methods. In the second stage, we begin by introducing semantic-aware augmentations, including structure-contextualized text augmentation and semantic-aware hyperedge dropping, to facilitate informative view generation. Subsequently, we propose a multi-scale contrastive loss with an $s$-walk-based subgraph-level objective to capture long-range dependencies. Extensive experiments on six real-world datasets validate the effectiveness of our proposed method.

CD300ld on pathologically activated neutrophils promotes tumor immune suppression by binding phosphatidylserine on CD8<sup>+</sup> T cells

✇Nature Cancer
By: Chaoxiong Wang · Peixuan Zheng · Anhui Wang · Xinrui Fan · Tianxiang Li · Jingjing Hong · Runze Wang · Chunyan Zhang · Chaogui Luo · Weiwei Yang · Dong Gao · Adili Salai · Min Luo · Zhigang Lu · Dianfan Li · Moubin Lin · Jiayin Peng · Guohui Li · Shuo Han · Yun Zhao
15 May 2026 at 08:00

Nature Cancer, Published online: 15 May 2026; doi:10.1038/s43018-026-01169-4

Zhao and colleagues show that CD300ld, upregulated in pathologically activated neutrophils, mediates contact-dependent suppression of cytotoxic CD8+ T cells by binding to phosphatidylserine, inhibiting antitumor immune responses.

An Orally Deliverable, Food-Compatible Lyophilized Recombinant Whole-Cell Catalyst for Alcohol-Associated Liver Injury

✇Omics in Gastric
By: Fan Li · Meng-Yue Zhang · Xiao-Le Shan · Cai-Yun Wang · Ying-Ying Wu · Shuang Li · Shi-Qiao Xu · Yi-Xuan Zhang
4 May 2026 at 18:00

Microorganisms. 2026 Mar 26;14(4):746. doi: 10.3390/microorganisms14040746.

ABSTRACT

Effective oral interventions for alcohol-induced metabolic stress and liver injury remain limited. Pre-absorptive gastrointestinal alcohol handling is gaining interest as a non-pharmacological strategy to reduce hepatic burden. In this study, we developed a formulation-integrated, food-compatible lyophilized recombinant whole-cell catalyst based on Escherichia coli Nissle 1917 engineered to express alcohol dehydrogenase and acetaldehyde dehydrogenase. Rather than focusing exclusively on strain-level genetic modification, the engineered cells were protected by lyophilization combined with a food-grade chitosan-alginate layer-by-layer coating, forming an artificial cell wall designed to enhance survivability during oral delivery. The formulation resisted simulated gastric acid, sodium taurocholate, and ethanol, retained enzymatic activity after storage, and demonstrated formulation stability. In alcohol-exposed mice, oral administration reduced blood ethanol and acetaldehyde levels, improved liver biochemical parameters, attenuated hepatic steatosis, and partially restored oxidative stress indicators. Integrated multi-omics analyses indicated coordinated gut-associated metabolic and inflammatory responses to alcohol and intervention, rather than a single dominant pathway. These findings provide hypothesis-generating evidence; causality remains to be established. Overall, this study demonstrates a proof-of-concept, food-compatible lyophilized recombinant whole-cell catalyst that integrates enzymatic function with formulation stability and gastrointestinal resilience, highlighting an applied, food-compatible microbial framework for exploring alcohol-related metabolic stress.

PMID:42075143 | PMC:PMC13119499 | DOI:10.3390/microorganisms14040746

Respiratory viral infections prime accelerated lung cancer growth

✇Cell
By: Wei Qian · Xiaoqin Wei · Andrew J. Barros · Xiangyu Ye · Haibo Zhang · Qing Yu · Samuel P. Young · Eric V. Yeatts · Yury Park · Chaofan Li · Sijie Hao · Gislane Almeida-Santos · Jinyi Tang · Harish Narasimhan · Nicole A. Kirk · Valeria Molinary · Ying Li · Li Li · Bimal N. Desai · Peter Chen · Kwon-Sik Park · Anny Xiaobo Zhou · Jeffrey M. Sturek · Wei Chen · In Su Cheon · Jie Sun
11 March 2026 at 08:00
Severe COVID-19 is associated with an increased subsequent risk of lung cancer. Viral pneumonia induces durable lung epigenetic imprinting that promotes tumor-supportive neutrophils and impairs T cell immunity, which is reversible with combined CXCR2 inhibition and PD-L1 blockade.
Received — 13 April 2026 ⏭

Asymmetric selection of a rice immune module and rebuild of disease resistance

✇Nature - Issue - nature.com science feeds
By: Hui Lin · Fudan Chen · Guanyun Cheng · Bingxiao Yan · Meng Yuan · Jie Qiu · Yiduo Lu · Mingzhe Suo · Ying Chen · Yijie Wang · Kaixuan Cui · Xiangyu Gong · Shasha Liu · Bofan Liu · Jiyun Liu · Jianjun Wang · Rongbai Li · Bizeng Mao · Jianlong Xu · Jong-Seong Jeon · Xuehui Huang · Bin Han · Dong-Lei Yang · Qifei Gao · Haiming Xu · Yiwen Deng · Gongyou Chen · Zuhua He
8 April 2026 at 08:00

Nature, Published online: 08 April 2026; doi:10.1038/s41586-026-10361-6

Stacking XA48-mediated effector-triggered immunity with XA21-mediated pattern-triggered immunity in Oryza sativa japonica reconstitutes the broad-spectrum resistance from wild rice.
Received — 8 April 2026 ⏭

TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Shichao Ma · Zhiyuan Ma · Ming Yang · Xiaofan Li · Xing Wu · Jintao Du · Yu Cheng · Weiqiang Wang · Qiliang Liu · Zhengyang Zhou · Yang Wang
7 April 2026 at 12:00
arXiv:2601.22776v2 Announce Type: replace Abstract: Multi-turn tool-integrated reasoning enables Large Language Models (LLMs) to solve complex tasks through iterative information retrieval. However, current reinforcement learning (RL) frameworks for search-augmented reasoning predominantly rely on sparse outcome-level rewards, leading to a "Double Homogenization Dilemma." This manifests as (1) Process homogenization, where the thinking, reasoning, and tooling involved in generation are ignored. (2) Intra-group homogenization, coarse-grained outcome rewards often lead to inefficiencies in intra-group advantage estimation with methods like Group Relative Policy Optimization (GRPO) during sampling. To address this, we propose Turn-level Stage-aware Policy Optimization (TSPO). TSPO introduces the First-Occurrence Latent Reward (FOLR) mechanism, allocating partial rewards to the step where the ground-truth answer first appears, thereby preserving process-level signals and increasing reward variance within groups without requiring external reward models or any annotations. Extensive experiments demonstrate that TSPO significantly outperforms state-of-the-art baselines, achieving average performance gains of 24% and 13.6% on Qwen2.5-3B and 7B models, respectively. Code is available at https://github.com/Flipped-May/TSPO.

Enhancing Foundation VLM Robustness to Missing Modality: Scalable Diffusion for Bi-directional Feature Restoration

✇cs.AI, q-bio.NC updates on arXiv.org
By: Wei Dai · Haoyu Wang · Honghao Chang · Lijun He · Fan Li · Jian Sun · Haixia Bi
7 April 2026 at 12:00
arXiv:2602.03151v2 Announce Type: replace Abstract: Vision Language Model (VLM) typically assume complete modality input during inference. However, their effectiveness drops sharply when certain modalities are unavailable or incomplete. Current research on missing modality primarily faces two dilemmas: Prompt-based methods struggle to restore missing yet indispensable features and degrade the generalizability of VLM. Imputation-based approaches, lacking effective guidance, are prone to generating semantically irrelevant noise. Restoring precise semantics while sustaining VLM's generalization remains challenging. Therefore, we propose a general missing modality restoration strategy in this paper. We introduce an enhanced diffusion model as a pluggable mid-stage training module to effectively restore missing features. Our strategy introduces two key innovations: (I) Dynamic Modality Gating, which adaptively leverages conditional features to guide the generation of semantically consistent features; (II) Cross-Modal Mutual Learning mechanism, which bridges the semantic spaces of the dual models to achieve bi-directional alignment. Notably, our strategy maintains the original integrity of the pre-trained VLM, requiring no fine-tuning of the backbone models while significantly boosting resilience to information loss. Zero-shot evaluations across benchmark datasets demonstrate that our approach consistently outperforms existing baselines, establishing it as a robust and scalable extension that ensures VLM reliability across diverse missing rates and conditions. Our code and models will be publicly available.
Received — 4 April 2026 ⏭

AeroTherm-GPT: A Verification-Centered LLM Framework for Thermal Protection System Engineering Workflows

✇cs.AI, q-bio.NC updates on arXiv.org
By: Chuhan Qiao · Jinglai Zheng · Jie Huang · Buyue Zhao · Fan Li · Haiming Huang
3 April 2026 at 12:00
arXiv:2604.01738v1 Announce Type: new Abstract: Integrating Large Language Models (LLMs) into hypersonic thermal protection system (TPS) design is bottlenecked by cascading constraint violations when generating executable simulation artifacts. General-purpose LLMs, treating generation as single-pass text completion, fail to satisfy the sequential, multi-gate constraints inherent in safety-critical engineering workflows. To address this, we propose AeroTherm-GPT, the first TPS-specialized LLM Agent, instantiated through a Constraint-Closed-Loop Generation (CCLG) framework. CCLG organizes TPS artifact generation as an iterative workflow comprising generation, validation, CDG-guided repair, execution, and audit. The Constraint Dependency Graph (CDG) encodes empirical co-resolution structure among constraint categories, directing repair toward upstream fault candidates based on lifecycle ordering priors and empirical co-resolution probabilities. This upstream-priority mechanism resolves multiple downstream violations per action, achieving a Root-Cause Fix Efficiency of 4.16 versus 1.76 for flat-checklist repair. Evaluated on HyTPS-Bench and validated against external benchmarks, AeroTherm-GPT achieves 88.7% End-to-End Success Rate (95% CI: 87.5-89.9), a gain of +12.5 pp over the matched non-CDG ablation baseline, without catastrophic forgetting on scientific reasoning and code generation tasks.

NCCL EP: Towards a Unified Expert Parallel Communication API for NCCL

✇cs.AI, q-bio.NC updates on arXiv.org
By: Amos Goldman (NVIDIA Corporation) · Nimrod Boker (NVIDIA Corporation) · Maayan Sheraizin (NVIDIA Corporation) · Nimrod Admoni (NVIDIA Corporation) · Artem Polyakov (NVIDIA Corporation) · Subhadeep Bhattacharya (NVIDIA Corporation) · Fan Yu (NVIDIA Corporation) · Kai Sun (NVIDIA Corporation) · Georgios Theodorakis (NVIDIA Corporation) · Hsin-Chun Yin (NVIDIA Corporation) · Peter-Jan Gootzen (NVIDIA Corporation) · Aamir Shafi (NVIDIA Corporation) · Assaf Ravid (NVIDIA Corporation) · Salvatore Di Girolamo (NVIDIA Corporation) · James Dinan (NVIDIA Corporation) · Xiaofan Li (NVIDIA Corporation) · Manjunath Gorentla Venkata (NVIDIA Corporation) · Gil Bloch (NVIDIA Corporation)
3 April 2026 at 12:00
arXiv:2603.13606v3 Announce Type: replace-cross Abstract: Mixture-of-Experts (MoE) architectures have become essential for scaling large language models, driving the development of specialized device-initiated communication libraries such as DeepEP, Hybrid-EP, and others. These libraries demonstrate the performance benefits of GPU-initiated RDMA for MoE dispatch and combine operations. This paper presents NCCL EP (Expert Parallelism), a ground-up MoE communication library built entirely on NCCL's Device API. NCCL EP provides unified ncclEpDispatch and ncclEpCombine primitives with both C and Python interfaces, supporting Low-Latency (LL) mode for inference decoding and High-Throughput (HT) mode for training and inference prefill. LL targets small batch sizes (1-128 tokens) using direct all-to-all RDMA+NVLink mesh connectivity with double-buffered communication for overlapping dispatch and combine phases. HT targets large batches (4096+ tokens) using hierarchical communication that aggregates tokens within NVLink domains before inter-node RDMA transmission. Both modes leverage Device API for both intra- and inter-node communications, taking advantage of its topology awareness and optimized GPU-initiated implementation. We evaluate NCCL EP on an H100-based cluster across multi-node configurations, demonstrating competitive LL kernel performance and presenting end-to-end results with vLLM integration. By building MoE communication natively within NCCL, NCCL EP provides a supported path for expert parallelism on current and emerging NVIDIA platforms.

Ferritin aggregation cell engager for CAR T avidity engineering against refractory leukemias

✇Cell
By: Feng Li · Yuxing Hu · Yan Wang · Xiao Zhang · Shaowei Qiu · Xiaoming Hu · Wenjing Wang · Shuang Wang · Peilin Guo · Cui Song · Huaiji Wang · Changhao He · Jie Sun · Di Yu · Shuhua Yi · Jun Wei · Feifan Liu · Caopei Zheng · Jianxiang Wang · Yuhua Li · Guanghui Ma · Wei Wei
9 March 2026 at 08:00
Li et al. developed a ferritin aggregation cell engager that helps CAR T cells better recognize and attack leukemia cells without re-engineering the CAR itself. This versatile platform overcomes antigen modulation and enables combination with chemotherapy.
Received — 26 March 2026 ⏭

Hijacking ERAD for targeted degradation of transmembrane proteins

✇Cell
By: Haikun Song · Wei Wang · Tingfang Mei · Huiwen Zheng · Keni Ning · Xiaofan Liu · Siulam Cheung · Zhiyuan Cao · Danqi Sheng · Xiaohan Mai · Haoran Zhu · Guankai Guo · Shuaimeng Liu · Rongkai Wei · Qian Wang · Yu Cao · Yu Ding · Yiyan Fei · Rui Liu · Motoyuki Hattori · Chunquan Sheng · Boxun Lu
19 March 2026 at 08:00
Development of an ERAD-hijacking technology overcomes the challenges of current targeted protein degradation approaches to achieve degradation of transmembrane proteins.

Sustained nitric oxide production by engineered <i>E. coli</i> remodels the tumor microenvironment and potentiates immunotherapy

✇Nature Biotechnology - Issue - nature.com science feeds
By: Shuyu Xu · Tianjiao Zhang · Yang Song · Mengxin Wang · Ruiqi Wu · Mofan Li · Haonan Wang · Xinxin Xie · Qingfeng Chen · Xiaotu Ma · Xiaolong Liang
18 March 2026 at 08:00

Nature Biotechnology, Published online: 18 March 2026; doi:10.1038/s41587-026-03054-y

Solid tumors are sensitized to anti‑PD‑L1 immunotherapy by engineered E. coli to produce nitric oxide.
Received — 17 March 2026 ⏭

FCMBench: The First Large-scale Financial Credit Multimodal Benchmark for Real-world Applications

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yehui Yang · Dalu Yang · Fangxin Shang · Wenshuo Zhou · Jie Ren · Yifan Liu · Haojun Fei · Qing Yang · Yanwu Xu · Tao Chen
16 March 2026 at 12:00
arXiv:2601.00150v3 Announce Type: replace-cross Abstract: FCMBench is the first large-scale and privacy-compliant multimodal benchmark for real-world financial credit applications, covering tasks and robustness challenges from domain specific workflows and constraints. The current version of FCMBench covers 26 certificate types, with 5198 privacy-compliant images and 13806 paired VQA samples. It evaluates models on Perception and Reasoning tasks under real-world Robustness interferences, including 3 foundational perception tasks, 4 credit-specific reasoning tasks demanding decision-oriented visual evidence interpretation, and 10 real-world challenges for rigorous robustness stress testing. Moreover, FCMBench offers privacy-compliant realism with minimal leakage risk through in-house scenario-aware captures of manually synthesized templates, without any publicly released images. We conduct extensive evaluations of 28 state-of-the-art vision-language models spanning 14 AI companies and research institutes. Among them, Gemini 3 Pro achieves the best F1 score as a commercial model (65.16), Kimi-K2.5 achieves the best score as an open-source baseline (60.58). The mean and the std. of all tested models is 44.8 and 10.3 respectively, indicating that FCMBench is non-trivial and provides strong resolution for separating modern vision-language model capabilities. Robustness evaluations reveal that even top-performing models experience notable performance degradation under the designed challenges. We have open-sourced this benchmark to advance AI research in the credit domain and provide a domain-specific task for real-world AI applications.
Received — 11 March 2026 ⏭

DARC: Disagreement-Aware Alignment via Risk-Constrained Decoding

✇cs.AI, q-bio.NC updates on arXiv.org
By: Mingxi Zou · Jiaxiang Chen · Junfan Li · Langzhang Liang · Qifan Wang · Xu Yinghui · Zenglin Xu
10 March 2026 at 12:00
arXiv:2603.08145v1 Announce Type: cross Abstract: Preference-based alignment methods (e.g., RLHF, DPO) typically optimize a single scalar objective, implicitly averaging over heterogeneous human preferences. In practice, systematic annotator and user-group disagreement makes mean-reward maximization brittle and susceptible to proxy over-optimization. We propose **Disagreement-Aware Alignment via Risk-Constrained Decoding (DARC)**, a retraining-free inference-time method that frames response selection as distributionally robust, risk-sensitive decision making. Given multiple preference samples or scalable disagreement proxies, DARC reranks candidates by maximizing a *KL-robust (entropic)* satisfaction objective, and provides simple deployment controls that cap or penalize the corresponding entropic risk premium relative to the mean, enabling explicit risk budgets without retraining. We provide theoretical characterization linking this decoding rule to principled pessimism and KL-based distributionally robust optimization. Experiments on alignment benchmarks show that DARC reduces disagreement and tail risk while maintaining competitive average quality under noisy, heterogeneous feedback.
❌