1 PyTorch 性能剖析（第 2 部分）：从 nn.Linear 到融合 MLP

2 PyTorch 性能剖析（第 1 部分）：torch.profiler 入门指南

3 Codex 正在推动 AI 模型的开源与训练流程

4 构建开放的未来——我们与 Google Cloud 的全新合作伙伴关系

5 使用 NVIDIA Isaac 构建医疗机器人：从仿真到部署

6 经同意的语音克隆

7 流式数据集：效率提升 100 倍

8 huggingface_hub v1.0：开源机器学习基础五周年回顾

9 LeRobot v0.4.0：全面提升开源机器人的学习能力

10 用开源模型强化你的 OCR 工作流

11 用 AI Sheets 解锁图像的力量

12 来自OpenAI gpt-oss的技巧，你🫵在transformers中也可以使用

13 ZeroGPU Spaces 加速实践：PyTorch Ahead-of-Time Compilation 全解析

14 用 Claude 和 Hugging Face 生成图像

1 ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimization

2 Argo-Bench: Evaluating Data Agents on Enterprise-Scale Workflows

3 Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video Generation

4 Decoding Looped Transformers Better for (Almost) Free

5 KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards

6 ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research

7 AgSpec: Pushing the Limits of Retrieval-Based Speculative Decoding in Coding Agent Pipelines

8 Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes

9 InterEvolve: Test-Time Evolution of Reward Programs for Humanoid Loco-Manipulation

10 AutoGUIWorld: Image Generators as Visual World Models for GUI Agent

11 ROWBench: Do Video Models Render What the Program Specifies?

12 World Observer: Joint Actor-Observer Generation for Persistent World Modeling

13 Beyond Memory: Harnessing Long-Horizon Agents with Explicit Belief States

14 Sharpening Tax in Post-Training

15 Hierarchical Continuous Diffusion Language Models

16 RPTune: Learned Context Curation for LLM Catalog Search

17 Omni-Embed-Mini: Binding Modalities Without Forgetting via Dense Distillation

18 Scaling and Distilling Text Embeddings for Better Diffusibility

19 OmniSeek: Native Tool Integration for Multi-turn Audio-Visual Reasoning

20 Fewer Tokens, Better Action: GPT-6 Astra Robot Agents with 14% Higher Success Rate but 65% Fewer Tokens

21 VTR-Bench: A Systematic Benchmark for Evaluating Visual Text Rendering in Video Generation

22 SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation

23 Generalization Is Stability, Not Accuracy: Multi-Axis Evaluation of LLMs

24 4Director: Controlling Video World Models with Rigid 3D Geometry

25 It Takes Workflows to Evolve Better Workflows

26 Keyword Harnesses Fail Open: A Cheap Diagnostic Ladder for Tool-Use Claims in Small Language Models

27 Pay for the Fault, Not the Flow: Label-Free In-Flow Multi-Agent Workflow Optimization

28 Before It Fades: Reinforcing Temporal Representations at Inference Time in VideoLLMs

29 Latent-Foresight: End-to-End Learning Predictable Representations for Latent World Models

30 OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction

31 LOCI: Spatial Linear Memory for Streaming World Models

32 PhysVista: Benchmarking Physical Intelligence in VLMs via a Perception-Reasoning-Assessment Loop

33 Multimodal Flow: Unified Flow Modeling of Language and Vision in Embedding Spaces

34 JevSpawn: Adaptive Agentic Inference through Compositional Action Spaces

35 Memorizon: Training World Models Beyond Their Context Window

36 Align Then Reason: A Multimodal Lip-Sync Judge for Dubbing

37 Smaller Models, Better Rejects: Preference Distillation Scaling

38 Make Sparse Rewards Count: Density-Aware Reward Aggregation for Multi-Reward RL

39 A Missing Piece for Trustworthy AI Reviewers: From Benchmarking Rhetorical Robustness to SciCore Review

40 GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis

41 PixelDense: Dense Prediction as Representation Alignment for Pixel Diffusion

42 EgoTools: Towards Tool-Centric Reasoning in Real-World Egocentric Videos

43 Video Generation Models: A Survey of Post-Training and Alignment

44 Benchmarking and Enhancing Skill-Level Memory for Partially Observable Robotic Manipulation

45 Beyond the Current Scene: Event-Referential Grasping with Active View Selection

46 SemanTok: Predictable Semantic Tokens for Efficient Autoregressive Video Generation

47 Architect-Ant: Editable Automatic Furnishing of Architectural Floor Plans

48 Personalized Image Generation with Reasoning and Reflection

49 AutoDataBench: A Data-centric Testbed for Accelerating Auto Research

50 DexPolicy: Scheduled Exploration for Trajectory-Guided Dexterous Manipulation

51 Honeycomb: Constant-Size Scene Memory Representation for Video World Models

52 Better Supervision Is Nearby: Neighborhood On-Policy Self-Distillation

53 FlexRouter: Learning Complementary Model Sets for Flexible LLM Routing

54 MemFold: Learning Compact Soft Memory for Long-Context Personalization via On-Policy Optimization

55 Explore Broadly, Reason Sharply: Push Small Models toward the Frontier via Sampling

56 Prefill-Free Cross-Family KV Cache Transfer for Heterogeneous Multi-Agent LLMs

57 Prompt2Skill: Unsupervised Skill Optimization From Natural Language Instructions

58 Rules to Tools: Executable Checks for LLM Agents in Scientific Computing

59 Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation

60 Predictive Credit: Measuring What Scientific Explanations Add to Experimental Forecasts

61 Pretrain Once, Route Anywhere: Towards a Foundation Model for LLM Routing

62 RLE-Bench: A Qualifying Exam for Coding Agents as Robot Learning Engineers

63 CorrGRPO: Correlation-Normalized GRPO for Multi-Reward Learning

64 Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation

65 E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Diffusion Language Models

66 Controlled Decoding Attacks on Black-Box LLMs

67 Devils in Question Relay: Source-Conditioned Relay Steering to Mitigate Hallucinations in Audio-visual Large Language Models

68 Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes It

69 OTRetarget: Joint Robot and Object Motion Retargeting via Optimal Transport

70 Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL

71 When Does Correction Become Repair? Mechanistic Auditing of Internal Interventions in Tool-Using LLMs

72 Learning What to Recall: Adaptive Multi-Cue Episodic Memory for World Models

73 Joint and Cross-Modal Video-Audio Generation and Editing: A Unified Formulation and Design Taxonomy

74 Does Native 3D Texture Generation Necessarily Require 3D Assets for Training?

75 Persona Dosing: Calibrated Activation Steering for Graded Trait Control

76 Agent Priors-guided Policy Learning

77 On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics

78 DataMagic: Authoring Data Videos through Declarative Multi-Agent Orchestration

79 Replacing Large Language Models with Jev Decision Models for Low-Latency Edge Service Orchestration

80 Do Audio LLMs Listen Before They Act? Diagnosing Acoustic-Context Gating in Voice Agents

81 OpenTumorBoard: A Real-World Benchmark of Multidisciplinary Tumor Board Discussion Trajectories

82 When Users Change Their Minds: Measuring and Repairing Intent Drift in LLM Agents

83 X-Tree: Tokenizing Reusable Experience for Efficient Agent Generalization

84 Decentralized Master-Mind: Joint Action Refinement through Iterative Intent Denoising in Multi-Agent Pathfinding

1 The Agent Said It Was Done. The Database Disagreed.

2 Open-sourcing AstaBrief, the fast report-generation model in Asta

3 AutoSynthData: Generating Training Data for Enterprise Agents

4 Open TTS Leaderboard: Scalable Evaluation for Multilingual Text-to-Speech and Voice Cloning

5 NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction

6 Getting the Source Right, Not Just the Fact: Source-Aware Verification for MCP Agents

7 Holo4: powering generalist computer-use agents

8 Accelerating vision-language models with LFM2.5-VL-DSpark

9 Jun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX community

10 Transformers now runs llama.cpp quants

11 How UK AISI and EvalEval Are Making Benchmark Results Reproducible

12 tokenizers v1: encode, decode and scaling, measured

13 Your Agent Aced the Task. Will It Do It Again?

14 Rebuilding AUTOMATIC1111 with Gradio Workflow

15 Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL

16 NeoMME: an efficient Multimodal-native and Multilingual Encoder

17 Training a coding model to paint watercolours with TRL and OpenEnv

18 Give Your Coding Agents a Memory You Own

19 Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps

20 BenchMIRT: What are LLM benchmarks actually measuring?

21 Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI

22 The Open ASR Leaderboard Adds Its First Global South Language

23 Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers

24 Granite 4.2 LLMs: How They're Built

25 Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original

26 Wire It, Run It, Deploy It: AI Workflows in Gradio

27 Measuring benchmark optimization in speech recognition

28 How Hugging Face Inference Endpoints, Jobs, and Buckets Power Search on Papers with Code

29 Up to 3.2x Faster Inference with LFM2.5-DSpark

30 How Much Memory Does Your Agent Actually Need?

31 Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers

32 Same Cluster, 33 Points More Utilization: What Changed Was the Order

33 State of Open Models: Summer 2026 Observations

34 Record, train, and deploy from one place with Strands Agents, LeRobot, and Hugging Face Storage Buckets

35 What We Learned by Reproducing 2,200 papers from ICML

36 Introducing OlmoEarth embeddings: Custom embedding exports from OlmoEarth Studio for downstream analysis

37 Thinking of ACE? We Can Do It with Fewer Tokens

38 Build Low-Latency Multilingual Voice Agents: Open Weights & Full Deployment Control with NVIDIA Magpie TTS

39 Making Knowledge Distillation Cheap Enough to Run at Scale

40 Meta is back with Muse Glimmer: local, agentic, multimodal, and open source

41 Baseten on Hugging Face Inference Providers 🔥

42 GPU Management: Why Idle GPUs Are the New Grounded Aircraft

43 NVIDIA Cosmos-H-Dreams: Bringing Real-Time Generative Simulation to Surgical Robotics

44 Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident

45 Bringing Nunchaku 4-bit Diffusion Inference to Diffusers

46 Grabette: an open system to record robot-manipulation data

47 Newer Models, Same Advantage

48 Security incident disclosure — July 2026

49 Model Routing Is Simple. Until It Isn’t.

50 Introducing Real World VoiceEQ: Measuring the human quality of voice AI

51 Welcome Inkling by Thinking Machines

52 Profiling in PyTorch (Part 3): Attention is all you profile

53 Native-speed vLLM transformers modeling backend

54 From Hugging Face to Amazon SageMaker Studio in one click

55 Hugging Face Models on Foundry Managed Compute

56 Run AI workloads on any cloud, store on Hugging Face: zero-egress storage with SkyPilot

57 LeRobot v0.6.0: Imagine, Evaluate, Improve

58 PRX Part 4: Our Data Strategy

59 🤗 Kernels: Major Updates

60 Hugging Face and Cerebras bring Gemma 4 to real-time voice AI

61 ScarfBench: Benchmarking AI Agents for Enterprise Java Framework Migration

62 Why Specialization Is Inevitable

63 Featuring Every Eval Ever Results on Hugging Face Model Pages

64 DiScoFormer: One transformer for density and score, across distributions

65 Run a vLLM Server on HF Jobs in One Command

66 Accelerating Transformers Fine-Tuning with NVIDIA NeMo AutoModel

67 Introducing the FFASR Leaderboard: Benchmarking ASR in the Real World

68 Experimenting with the proposed Cross-Origin Storage API in Transformers.js

69 Shipping huggingface_hub every week with AI, open tools, and a human in the loop

70 PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters

71 We got local models to triage the OpenClaw repo for FREE!*

72 MosaicLeaks: Can your research agent keep a secret?

73 Is it agentic enough? Benchmarking open models on your own tooling

74 Beyond LoRA: Can you beat the most popular fine-tuning technique?

75 From the Hugging Face Hub to robot hardware with Strands Agents and LeRobot

76 GLM-5.2: Built for Long-Horizon Tasks

77 Agentic Resource Discovery: Let agents search

78 Profiling in PyTorch (Part 2): From nn.Linear to a Fused MLP

79 How an Agent Built a 3D Paris Gallery by Chaining Two Hugging Face Spaces

80 Migrating Your GitHub CI to Hugging Face Jobs

81 The Open Source Community is backing OpenEnv for Agentic RL

82 Nemotron 3.5 Content Safety: Customizable Multimodal Safety for Global Enterprise AI

83 Designing the hf CLI as an agent-optimized way to work with the Hub

84 Direct Preference Optimization Beyond Chatbots

85 Adding MCP Tools to Reachy Mini

86 Holo3.1: Fast & Local Computer Use Agents

87 Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains

88 Beyond LLMs: Why Scalable Enterprise AI Adoption Depends on Agent Logic

89 Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler

90 Shipping a Trillion Parameters With a Hub Bucket: Delta Weight Sync in TRL

91 Reachy Mini goes fully local

92 Harness, Scaffold, and the AI Agent Terms Worth Getting Right

93 OlmoEarth v1.1: A more efficient family of Earth observation models

94 Introducing the Ettin Reranker Family

95 PaddleOCR 3.5: Running OCR and Document Parsing Tasks with a Transformers Backend

96 Granite Embedding Multilingual R2: Open Apache 2.0 Multilingual Embeddings with 32K Context — Best Sub-100M Retrieval Quality

97 Unlocking asynchronicity in continuous batching

98 Building Blocks for Foundation Model Training and Inference on AWS

99 vLLM V0 to V1: Correctness Before Corrections in RL

100 Adding Benchmaxxer Repellant to the Open ASR Leaderboard