HN hacker27369 2/12/2026

GodHands – Deterministic Desktop Automation via MCP

GodHands: MCP server for deterministic desktop automation treating apps as structured APIs instead of vision-based screenshots for high-reliability AI agents.

HF Xirui Li, Ming Li, Tianyi Zhou 2/12/2026

What does RL improve for Visual Reasoning? A Frankenstein-Style Analysis

Frankenstein-style analysis framework isolating specific visual reasoning improvements from RL versus supervised fine-tuning in vision-language models.

HN fabienpenso 2/12/2026

Moltis: Rust based AI assistant with memory, tools, and self-extending skills

Moltis: Rust-based AI agent with memory, tools, and runtime self-extending skills. Single binary with web UI for local, trustworthy AI assistant.

HF Tunyu Zhang, Xinxi Zhang, Ligong Han, Haizhou Shi, Xiaoxiao He, Zhuowei Li, Hao Wang, Kai Xu, Akash Srivastava, Hao Wang, Vladimir Pavlovic, Dimitris N. Metaxas 2/12/2026

T3D: Few-Step Diffusion Language Models via Trajectory Self-Distillation with Direct Discriminative Optimization

T3D proposes trajectory self-distillation framework to enable fast parallel token decoding in diffusion LLMs with fewer refinement steps while maintaining generation quality.

HF Dianyi Wang, Ruihang Li, Feng Han, Chaofan Ma, Wei Song, Siyuan Wang, Yibin Wang, Yi Xin, Hongjian Liu, Zhixiong Zhang, Shengyuan Ding, Tianhang Wang, Zhenglin Cheng, Tao Lin, Cheng Jin, Kaicheng Yu, Jingjing Chen, Wenjie Wang, Zhongyu Wei, Jiaqi Wang 2/12/2026

DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing

DeepGen 1.0 is a lightweight 5B unified model for image generation and editing using Stacked Channel Bridging, achieving competitive performance to larger models with reduced deployment costs.

HF Mathieu Sibue, Andres Muñoz Garza, Samuel Mensah, Pranav Shetty, Zhiqiang Ma, Xiaomo Liu, Manuela Veloso 2/12/2026

ExStrucTiny: A Benchmark for Schema-Variable Structured Information Extraction from Document Images

ExStrucTiny benchmark evaluates VLM performance on schema-variable structured information extraction from diverse enterprise documents with flexible schemas.

HF Xiaohan He, Shiyang Feng, Songtao Huang, Lei Bai, Bin Wang, Bo Zhang 2/12/2026

Sci-CoE: Co-evolving Scientific Reasoning LLMs via Geometric Consensus with Sparse Supervision

Sci-CoE framework enables LLMs to self-improve on scientific reasoning through co-evolution as both solver and verifier with geometric consensus mechanisms.

HF Sicheng Feng, Zigeng Chen, Xinyin Ma, Gongfan Fang, Xinchao Wang 2/12/2026

dVoting: Fast Voting for dLLMs

dVoting fast voting technique for diffusion LLMs enabling parallel test-time scaling for improved reasoning performance.

HF Wenkai Yang, Weijie Liu, Ruobing Xie, Kai Yang, Saiyong Yang, Yankai Lin 2/12/2026

Learning beyond Teacher: Generalized On-Policy Distillation with Reward Extrapolation

Theoretical and empirical analysis of on-policy distillation as dense KL-constrained RL, proposing generalized reward extrapolation.

HF Pinyi Zhang, Ting-En Lin, Yuchuan Wu, Jingyang Chen, Zongqi Wang, Hua Yang, Ze Xu, Fei Huang, Kai Zhang, Yongbin Li 2/12/2026

P-GenRM: Personalized Generative Reward Model with Test-time User-based Scaling

P-GenRM enables personalized LLM alignment through scenario-specific reward models with test-time user-based scaling, addressing generalization to new users with limited feedback.

HF Xiaoyuan Liu, Tian Liang, Dongyang Ma, Deyu Zhou, Haitao Mi, Pinjia He, Yan Wang 2/12/2026

The Pensieve Paradigm: Stateful Language Models Mastering Their Own Context

StateLM foundation model framework giving LLMs agency to manage their own context and memory via database operations.

HF GigaBrain Team, Boyuan Wang, Chaojun Ni, Guan Huang, Guosheng Zhao, Hao Li, Jie Li, Jindi Lv, Jingyu Liu, Lv Feng, Mingming Yu, Peng Li, Qiuping Deng, Tianze Liu, Xinyu Zhou, Xinze Chen, Xiaofeng Wang, Yang Wang, Yifan Li, Yifei Nie, Yilong Li, Yukun Zhou, Yun Ye, Zhichao Liu, Zheng Zhu 2/12/2026

GigaBrain-0.5M*: a VLA That Learns From World Model-Based Reinforcement Learning

GigaBrain-0.5M VLA model trained via world model-based reinforcement learning for improved multi-step action prediction.

HF Bo Zhang, Jiaxuan Guo, Lijun Li, Dongrui Liu, Sujin Chen, Guanxu Chen, Zhijie Zheng, Qihao Lin, Lewen Yan, Chen Qian, Yijin Zhou, Yuyao Wu, Shaoxiong Guo, Tianyi Du, Jingyi Yang, Xuhao Hu, Ziqi Miao, Xiaoya Lu, Jing Shao, Xia Hu 2/12/2026

DeepSight: An All-in-One LM Safety Toolkit

DeepSight is a unified toolkit for LLM/MLLM safety covering workflow, evaluation, diagnosis, and alignment with integrated explainability and risk scenario grounding capabilities.

HF Xinyu Yang, Chenlong Deng, Tongyu Wen, Binyu Xie, Zhicheng Dou 2/12/2026

LawThinker: A Deep Research Legal Agent in Dynamic Environments

LawThinker autonomous legal research agent using Explore-Verify-Memorize strategy with intermediate step verification in dynamic environments.

HF Xin Xu, Clive Bai, Kai Yang, Tianhao Chen, Yangkun Chen, Weijie Liu, Hao Chen, Yang Wang, Saiyong Yang, Can Yang 2/12/2026

Composition-RL: Compose Your Verifiable Prompts for Reinforcement Learning of Large Language Models

Composition-RL optimizes RLVR training by composing verifiable prompts to balance hard and easy examples, mitigating ineffective data and enabling better prompt dataset expansion.

HF Romain Froger, Pierre Andrews, Matteo Bettini, Amar Budhiraja, Ricardo Silveira Cabral, Virginie Do, Emilien Garreau, Jean-Baptiste Gaya, Hugo Laurençon, Maxime Lecanu, Kunal Malkan, Dheeraj Mekala, Pierre Ménard, Gerard Moreno-Torres Bertran, Ulyana Piterbarg, Mikhail Plekhanov, Mathieu Rita, Andrey Rusakov, Vladislav Vorotilov, Mengjue Wang, Ian Yu, Amine Benhalloum, Grégoire Mialon, Thomas Scialom 2/12/2026

Gaia2: Benchmarking LLM Agents on Dynamic and Asynchronous Environments

Gaia2 benchmark for evaluating LLM agents in realistic, asynchronous, dynamic environments with temporal constraints and collaboration.

HF Łukasz Staniszewski, Katarzyna Zaleska, Mateusz Modrzejewski, Kamil Deja 2/12/2026

TADA! Tuning Audio Diffusion Models through Activation Steering

TADA: activation steering technique for audio diffusion models to control semantic musical concepts through shared attention layers.

HF Nenad Tomašev, Matija Franklin, Simon Osindero 2/12/2026

Intelligent AI Delegation

Adaptive framework for intelligent AI agent delegation across decomposed sub-tasks with dynamic adaptation to environmental changes and failure handling.

HF Lai Wei, Liangbo He, Jun Lan, Lingzhong Dong, Yutong Cai, Siyuan Li, Huijia Zhu, Weiqiang Wang, Linghe Kong, Yue Wang, Zhuosheng Zhang, Weiran Huang 2/12/2026

Zooming without Zooming: Region-to-Image Distillation for Fine-Grained Multimodal Perception

Region-to-Image Distillation: reduces latency in multimodal LLMs' fine-grained perception by distilling zooming behavior into inference-time efficiency.

HF Hongbo Zhang, Yue Yang, Jianhao Yan, Guangsheng Bao, Yue Zhang, Yue Zhang 2/12/2026

Detecting RLVR Training Data via Structural Convergence of Reasoning

Detection method for identifying RLVR training data contamination via structural convergence signatures in reasoning trajectories, addressing benchmark contamination concerns.

BL 2/12/2026

Introducing GPT-5.3-Codex-Spark

GPT-5.3-Codex-Spark real-time coding model with 15x faster generation, 128k context, available in research preview.

HF Zhenghuang Wu, Kang Chen, Zeyu Zhang, Hao Tang 2/12/2026

Light4D: Training-Free Extreme Viewpoint 4D Video Relighting

Light4D: training-free framework for 4D video relighting under extreme viewpoints using diffusion models with temporal consistency.

HF MiniCPM Team, Wenhao An, Yingfa Chen, Yewei Fang, Jiayi Li, Xin Li, Yaohui Li, Yishan Li, Yuxuan Li, Biyuan Lin, Chuan Liu, Hezi Liu, Siyuan Liu, Hongya Lyu, Yinxu Pan, Shixin Ren, Xingyu Shen, Zhou Su, Haojun Sun, Yangang Sun, Zhen Leng Thai, Xin Tian, Rui Wang, Xiaorong Wang, Yudong Wang, Bo Wu, Xiaoyue Xu, Dong Xu, Shuaikang Xue, Jiawei Yang, Bowen Zhang, Jinqian Zhang, Letian Zhang, Shengnan Zhang, Xinyu Zhang, Xinyuan Zhang, Zhu Zhang, Hengyu Zhao, Jiacheng Zhao, Jie Zhou, Zihan Zhou, Shuo Wang, Chaojun Xiao, Xu Han, Zhiyuan Liu, Maosong Sun 2/12/2026

MiniCPM-SALA: Hybridizing Sparse and Linear Attention for Efficient Long-Context Modeling

MiniCPM-SALA hybrid sparse-linear attention architecture for efficient long-context LLM processing in 9B parameter model.

HF Yi Zhang, Yunshuang Wang, Zeyu Zhang, Hao Tang 2/12/2026

Code2Worlds: Empowering Coding LLMs for 4D World Generation

Code2Worlds: extends coding LLMs to 4D world generation with dynamic physics simulation by addressing multi-scale context and semantic-physical execution gaps.

HF Futing Wang, Jianhao Yan, Yun Luo, Ganqu Cui, Zhi Wang, Xiaoye Qu, Yue Zhang, Yu Cheng, Tao Lin 2/12/2026

Think Longer to Explore Deeper: Learn to Explore In-Context via Length-Incentivized Reinforcement Learning

Framework enabling LLMs to perform in-context exploration through length-incentivized RL, allowing models to generate, verify, and refine multiple reasoning hypotheses within continuous context.

HF Matteo Nulli, Vladimir Orshulevich, Tala Bazazo, Christian Herold, Michael Kozielski, Marcin Mazur, Szymon Tuzel, Cees G. M. Snoek, Seyyed Hadi Hashemi, Omar Javed, Yannick Versley, Shahram Khadivi 2/12/2026

Adapting Vision-Language Models for E-commerce Understanding at Scale

Large-scale study on adapting general-purpose VLMs to e-commerce attribute understanding while preserving generalizability across multi-image noisy product data.

HF Haolei Bai, Lingcheng Kong, Xueyi Chen, Jianmian Wang, Zhiqiang Tao, Huan Wang 2/12/2026

DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels

Diffusion language models tailored for CUDA kernel code generation, addressing data scarcity and specialization challenges through parallel token generation approach.

HF Xin Xu, Tong Yu, Xiang Chen, Haoliang Wang, Julian McAuley, Saayan Mitra 2/12/2026

ThinkRouter: Efficient Reasoning via Routing Thinking between Latent and Discrete Spaces

ThinkRouter framework routing reasoning between latent and discrete spaces to improve efficiency based on model confidence dynamics.

HF Changti Wu, Jiahuai Mao, Yuzhuo Miao, Shijie Lian, Bin Yu, Xiaopeng Lin, Cong Huang, Lei Zhang, Kai Chen 2/12/2026

ScalSelect: Scalable Training-Free Multimodal Data Selection for Efficient Visual Instruction Tuning

ScalSelect training-free data selection method for efficient visual instruction tuning of vision-language models.

HF Zedong Chu, Shichao Xie, Xiaolong Wu, Yanfen Shen, Minghua Luo, Zhengbo Wang, Fei Liu, Xiaoxu Leng, Junjun Hu, Mingyang Yin, Jia Lu, Yingnan Guo, Kai Yang, Jiawei Han, Xu Chen, Yanqing Zhu, Yuxiang Zhao, Xin Liu, Yirong Yang, Ye He, Jiahang Wang, Yang Cai, Tianlin Zhang, Li Gao, Liu Liu, Mingchao Sun, Fan Jiang, Chiyu Wang, Zhicheng Liu, Hongyu Pan, Honglin Han, Zhining Gu, Kuan Yang, Jianfang Zhang, Di Jing, Zihao Guan, Wei Guo, Guoqing Liu, Di Yang, Xiangpo Yang, Menglin Yang, Hongguang Xing, Weiguo Li, Mu Xu 2/12/2026

ABot-N0: Technical Report on the VLA Foundation Model for Versatile Embodied Navigation

ABot-N0 unified Vision-Language-Action foundation model for embodied robot navigation across five core tasks using hierarchical brain-action architecture.

HF Jinrui Zhang, Chaodong Xiao, Aoqi Wu, Xindong Zhang, Lei Zhang 2/12/2026

Pretraining A Large Language Model using Distributed GPUs: A Memory-Efficient Decentralized Paradigm

SPES enables memory-efficient decentralized LLM pretraining using mixture-of-experts and distributed GPUs without full model replication on each node.

HF Hanbing Liu, Chunhao Tian, Nan An, Ziyuan Wang, Pinyan Lu, Changyuan Yu, Qi Qi 2/12/2026

Budget-Constrained Agentic Large Language Models: Intention-Based Planning for Costly Tool Use

INTENT framework for budget-constrained LLM agents solving multi-step tasks under monetary constraints via intention-based planning.

HF David Wan, Han Wang, Ziyang Wang, Elias Stengel-Eskin, Hyunji Lee, Mohit Bansal 2/12/2026

Multimodal Fact-Level Attribution for Verifiable Reasoning

MuRGAt benchmark evaluates multimodal LLM attribution and factual grounding across complex reasoning tasks involving multiple modalities and information sources.

HF Yejin Kim, Wilbert Pumacay, Omar Rayyan, Max Argus, Winson Han, Eli VanderBilt, Jordi Salvador, Abhay Deshpande, Rose Hendrix, Snehal Jauhri, Shuo Liu, Nur Muhammad Mahi Shafiullah, Maya Guru, Ainaz Eftekhar, Karen Farley, Donovan Clay, Jiafei Duan, Arjun Guru, Piper Wolters, Alvaro Herrasti, Ying-Chun Lee, Georgia Chalvatzaki, Yuchen Cui, Ali Farhadi, Dieter Fox, Ranjay Krishna 2/11/2026

MolmoSpaces: A Large-Scale Open Ecosystem for Robot Navigation and Manipulation

MolmoSpaces open ecosystem for large-scale benchmarking of robot navigation and manipulation policies with diverse scenarios.

HF Wayne Chi, Yixiong Fang, Arnav Yayavaram, Siddharth Yayavaram, Seth Karten, Qiuhong Anna Wei, Runkun Chen, Alexander Wang, Valerie Chen, Ameet Talwalkar, Chris Donahue 2/11/2026

GameDevBench: Evaluating Agentic Capabilities Through Game Development

GameDevBench evaluation framework tests multimodal AI agent capabilities on game development tasks combining code, shaders, sprites, and animations.

HF Yandan Yang, Shuang Zeng, Tong Lin, Xinyuan Chang, Dekang Qi, Junjin Xiao, Haoyun Liu, Ronghan Chen, Yuzhi Chen, Dongjie Huo, Feng Xiong, Xing Wei, Zhiheng Ma, Mu Xu 2/11/2026

ABot-M0: VLA Foundation Model for Robotic Manipulation with Action Manifold Learning

ABot-M0: VLA framework with action manifold learning for robotic manipulation, includes data curation pipeline for heterogeneous embodiment data.

HF Yitian Gong, Kuangwei Chen, Zhaoye Fei, Xiaogui Yang, Ke Chen, Yang Wang, Kexin Huang, Mingshu Chen, Ruixiao Li, Qingyuan Cheng, Shimin Li, Xipeng Qiu 2/11/2026

MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models

MOSS-Audio-Tokenizer proposes end-to-end discrete audio tokenization using homogeneous architectures for improved reconstruction and scaling in audio foundation models.

YT DeepLearningAI 2/11/2026

Use A2A to connect agents across different frameworks and teams

A2A framework enables connecting AI agents across different frameworks and teams for interoperability.

HF Jiahao Yuan, Yike Xu, Jinyong Wen, Baokun Wang, Yang Chen, Xiaotong Lin, Wuliang Huang, Ziyi Gao, Xing Fu, Yu Cheng, Weiqiang Wang 2/11/2026

How Do Decoder-Only LLMs Perceive Users? Rethinking Attention Masking for User Representation Learning

Research study on attention masking strategies in decoder-only LLMs for user representation learning using contrastive learning on large-scale behavioral data.

HF Guangzhi Xiong, Sanchit Sinha, Aidong Zhang 2/11/2026

Neural Additive Experts: Context-Gated Experts for Controllable Model Additivity

Neural Additive Experts framework balancing interpretability and accuracy in generalized additive models through feature interaction gating.

HF Chenhao Zhang, Yazhe Niu, Hongsheng Li 2/11/2026

MetaphorStar: Image Metaphor Understanding and Reasoning with End-to-End Visual Reinforcement Learning

MetaphorStar uses end-to-end visual RL to improve MLLM understanding of metaphorical content in images, enabling multi-hop reasoning and cultural context awareness.

HF Zhongzhi Li, Xuansheng Wu, Yijiang Li, Lijie Hu, Ninghao Liu 2/11/2026

Less is Enough: Synthesizing Diverse Data in Feature Space of LLMs

Feature Activation Coverage (FAC): measures post-training data diversity in LLM feature space for more effective downstream task performance.

HF Zhiling Yan, Dingjie Song, Zhe Fang, Yisheng Ji, Xiang Li, Quanzheng Li, Lichao Sun 2/10/2026

LiveMedBench: A Contamination-Free Medical Benchmark for LLMs with Automated Rubric Evaluation

Contamination-free medical benchmark for evaluating LLMs with automated rubric evaluation, addressing data leakage and temporal misalignment in clinical settings.

HF Weihao Liu, Dehai Min, Lu Cheng 2/10/2026

Latent Thoughts Tuning: Bridging Context and Reasoning with Fused Information in Latent Tokens

Method for reasoning in continuous latent space rather than discrete tokens, addressing feature collapse in latent reasoning paradigms for LLMs.

HF Chenxu Wang, Chaozhuo Li, Songyang Liu, Zejian Chen, Jinyu Hou, Ji Qi, Rui Li, Litian Zhang, Qiwei Ye, Zheng Liu, Xu Chen, Xi Zhang, Philip S. Yu 2/10/2026

The Devil Behind Moltbook: Anthropic Safety is Always Vanishing in Self-Evolving AI Societies

Theoretical and empirical analysis of safety alignment challenges in self-evolving multi-agent LLM systems; identifies self-evolution trilemma.

HF Checheng Yu, Chonghao Sima, Gangcheng Jiang, Hai Zhang, Haoguang Mai, Hongyang Li, Huijie Wang, Jin Chen, Kaiyang Wu, Li Chen, Lirui Zhao, Modi Shi, Ping Luo, Qingwen Bu, Shijia Peng, Tianyu Li, Yibo Yuan 2/9/2026

χ_{0}: Resource-Aware Robust Manipulation via Taming Distributional Inconsistencies

χ0 identifies distributional shift across human demonstrations, policy inductive bias, and test-time execution as robustness bottleneck in robotic manipulation, proposing alignment approach.

HF Suraj Ranganath, Atharv Ramesh 2/9/2026

StealthRL: Reinforcement Learning Paraphrase Attacks for Multi-Detector Evasion of AI-Text Detectors

StealthRL reinforcement learning framework stress-tests AI text detector robustness via adversarial paraphrasing using GRPO and LoRA adapters.

HF Feilong Tang, Xiang An, Yunyao Yan, Yin Xie, Bin Qin, Kaicheng Yang, Yifei Shen, Yuanhan Zhang, Chunyuan Li, Shikun Feng, Changrui Chen, Huajie Tan, Ming Hu, Manyuan Zhang, Bo Li, Ziyong Feng, Ziwei Liu, Zongyuan Ge, Jiankang Deng 2/9/2026

OneVision-Encoder: Codec-Aligned Sparsity as a Foundational Principle for Multimodal Intelligence

OneVision-Encoder: codec-aligned sparsity principle for multimodal architectures that process sparse discriminative information efficiently.

HF Konstantinos Mitsides, Maxence Faldor, Antoine Cully 2/9/2026

Dreaming in Code for Curriculum Learning in Open-Ended Worlds

Curriculum learning approach using code generation for agents to progressively learn in open-ended environments with foundation models.

HF Zhenyuan Zhang, Xianzhang Jia, Zhiqin Yang, Zhenbo Song, Wei Xue, Sirui Han, Yike Guo 2/8/2026

MemFly: On-the-Fly Memory Optimization via Information Bottleneck

MemFly: memory optimization framework using information bottleneck principles for LLM agents to balance compression and retrieval precision in long-term memory.