1 Codex 正在推动 AI 模型的开源与训练流程

2 构建开放的未来——我们与 Google Cloud 的全新合作伙伴关系

3 使用 NVIDIA Isaac 构建医疗机器人：从仿真到部署

4 经同意的语音克隆

5 流式数据集：效率提升 100 倍

6 huggingface_hub v1.0：开源机器学习基础五周年回顾

7 LeRobot v0.4.0：全面提升开源机器人的学习能力

8 用开源模型强化你的 OCR 工作流

9 用 AI Sheets 解锁图像的力量

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

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

12 欢迎 GPT OSS —— 来自 OpenAI 的全新开放模型家族！

13 SmolLM3: smol, multilingual, long-context reasoner

14 SmolVLA: Efficient Vision-Language-Action Model trained on Lerobot Community Data

1 Tapered Language Models

2 Dense Reward for Multi-View 3D Reasoning with Global Maps and Local Views

3 MeshFlow: Mesh Generation with Equivariant Flow Matching

4 KaLM-Reranker-V1: Fast but Not Late Interaction for Compressed Document Reranking

5 UniverSat: Resolution- and Modality-Agnostic Transformers for Earth Observation

6 Tmax: A simple recipe for terminal agents

7 Causal Discovery in the Era of Agents

8 Self-Compacting Language Model Agents

9 Training Open Models for Agentic Phone Use

10 Foresight: Failure Detection for Long-Horizon Robotic Manipulation with Action-Conditioned World Model Latents

11 Unlimited OCR Works

12 CLI-Universe: Towards Verifiable Task Synthesis Engine for Terminal Agents

13 HAKARI-Bench: A Lightweight Benchmark for Comparing Retrieval Architectures and Efficiency Settings under Unified Conditions

14 Safe Few-Step Generation via Velocity Editing

15 EnterpriseClawBench: Benchmarking Agents from Real Workplace Sessions

16 PolicyTrim: Boosting Intrinsic Policy Efficiency of Vision-Language-Action Models

17 PlanBench-XL: Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool Ecosystems

18 BioMatrix: Towards a Comprehensive Biological Foundation Model Spanning the Modality Matrix of Sequences, Structures, and Language

19 Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding

20 Improving Text-to-Music Generation with Human Preference Rewards

21 CalVerT: Augmenting Agents with Calibrated Verifier Telemetry Improves Action and Learning in Knowledge-Intensive Tasks

22 Counsel: A Meta-Evaluation Dataset for Agentic Tasks

23 PoLAR: Factorizing Extent and Mode in Latent Actions for Robot Policy Learning

24 EvoEmbedding: Evolvable Representations for Long-Context Retrieval and Agentic Memory

25 DataClaw0: Agentic Tailoring Multimodal Data from Raw Streams

26 Connect the Dots: Training LLMs for Long-Lifecycle Agents with Cross-Domain Generalization Via Reinforcement Learning

27 HydraHead: From Head-Level Functional Heterogeneity to Specialized Attention Hybridization

28 World Action Models: A Survey

29 Manifold Bandits: Bayesian Curriculum Learning over the Latent Geometry of Large Language Models

30 Deep Research in Physical Sciences: A Multi-Agent Framework and Comprehensive Benchmark

31 OpenRath: Session-Centered Runtime State for Agent Systems

32 Learning from Your Own Mistakes: Constructing Learnable Micro-Reflective Trajectories for Self-Distillation

33 FastMix: Fast Data Mixture Optimization via Gradient Descent

34 DailyReport: An Open-ended Benchmark for Evaluating Search Agents on Daily Search Tasks

35 Notes2Skills: From Lab Notebooks to Certainty-Aware Scientific Agent Skills

36 Exploring the Design Space of Reward Backpropagation for Flow Matching

37 SkillHarness: Harnessing Safe Skills for Computer-Use Agents

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

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

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

4 MosaicLeaks: Can your research agent keep a secret?

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

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

7 MolmoMotion: Language-guided 3D motion forecasting

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

9 GLM-5.2: Built for Long-Horizon Tasks

10 Agentic Resource Discovery: Let agents search

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

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

13 Migrating Your GitHub CI to Hugging Face Jobs

14 The Open Source Community is backing OpenEnv for Agentic RL

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

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

17 Direct Preference Optimization Beyond Chatbots

18 Adding MCP Tools to Reachy Mini

19 Holo3.1: Fast & Local Computer Use Agents

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

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

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

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

24 Reachy Mini goes fully local

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

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

27 Introducing the Ettin Reranker Family

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

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

30 Unlocking asynchronicity in continuous batching

31 Building Blocks for Foundation Model Training and Inference on AWS

32 vLLM V0 to V1: Correctness Before Corrections in RL

33 Adding Benchmaxxer Repellant to the Open ASR Leaderboard

34 Granite 4.1 LLMs: How They’re Built

35 DeepInfra on Hugging Face Inference Providers 🔥

36 Introducing NVIDIA Nemotron 3 Nano Omni: Long-Context Multimodal Intelligence for Documents, Audio and Video Agents

37 How to build scalable web apps with OpenAI's Privacy Filter

38 DeepSeek-V4: a million-token context that agents can actually use

39 How to Use Transformers.js in a Chrome Extension

40 QIMMA قِمّة ⛰: A Quality-First Arabic LLM Leaderboard

41 AI and the Future of Cybersecurity: Why Openness Matters

42 Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers

43 The PR you would have opened yourself

44 Ecom-RLVE: Adaptive Verifiable Environments for E-Commerce Conversational Agents

45 Inside VAKRA: Reasoning, Tool Use, and Failure Modes of Agents

46 Meet HoloTab by HCompany. Your AI browser companion.

47 Multimodal Embedding & Reranker Models with Sentence Transformers

48 Waypoint-1.5: Higher-Fidelity Interactive Worlds for Everyday GPUs

49 Safetensors is Joining the PyTorch Foundation

50 Welcome Gemma 4: Frontier multimodal intelligence on device

51 Falcon Perception

52 Any Custom Frontend with Gradio's Backend

53 Granite 4.0 3B Vision: Compact Multimodal Intelligence for Enterprise Documents

54 Training mRNA Language Models Across 25 Species for $165

55 TRL v1.0: Post-Training Library Built to Move with the Field

56 Liberate your OpenClaw

57 A New Framework for Evaluating Voice Agents (EVA)

58 Build a Domain-Specific Embedding Model in Under a Day

59 State of Open Source on Hugging Face: Spring 2026

60 Holotron-12B - High Throughput Computer Use Agent

61 Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries

62 Introducing Storage Buckets on the Hugging Face Hub

63 Ulysses Sequence Parallelism: Training with Million-Token Contexts

64 LeRobot v0.5.0: Scaling Every Dimension

65 Bringing Robotics AI to Embedded Platforms: Dataset Recording, VLA Fine‑Tuning, and On‑Device Optimizations

66 Introducing Modular Diffusers - Composable Building Blocks for Diffusion Pipelines

67 PRX Part 3 — Training a Text-to-Image Model in 24h!

68 Mixture of Experts (MoEs) in Transformers

69 Train AI models with Unsloth and Hugging Face Jobs for FREE

70 GGML and llama.cpp join HF to ensure the long-term progress of Local AI

71 IBM and UC Berkeley Diagnose Why Enterprise Agents Fail Using IT-Bench and MAST

72 One-Shot Any Web App with Gradio's gr.HTML

73 Custom Kernels for All from Codex and Claude

74 OpenEnv in Practice: Evaluating Tool-Using Agents in Real-World Environments

75 Transformers.js v4: Now Available on NPM!

76 Introducing SyGra Studio

77 Community Evals: Because we're done trusting black-box leaderboards over the community

78 H Company's new Holo2 model takes the lead in UI Localization

79 The Future of the Global Open-Source AI Ecosystem: From DeepSeek to AI+

80 Training Design for Text-to-Image Models: Lessons from Ablations

81 Introducing Daggr: Chain apps programmatically, inspect visually

82 We Got Claude to Build CUDA Kernels and teach open models!

83 Architectural Choices in China's Open-Source AI Ecosystem: Building Beyond DeepSeek 

84 Alyah ⭐️: Toward Robust Evaluation of Emirati Dialect Capabilities in Arabic LLMs

85 Unlocking Agentic RL Training for GPT-OSS: A Practical Retrospective

86 AssetOpsBench: Bridging the Gap Between AI Agent Benchmarks and Industrial Reality

87 One Year Since the “DeepSeek Moment”

88 Differential Transformer V2

89 Introducing Waypoint-1: Real-time interactive video diffusion from Overworld

90 Open Responses: What you need to know

91 NVIDIA Cosmos Reason 2 Brings Advanced Reasoning To Physical AI

92 Introducing Falcon-H1-Arabic: Pushing the Boundaries of Arabic Language AI with Hybrid Architecture

93 NVIDIA brings agents to life with DGX Spark and Reachy Mini

94 AprielGuard: A Guardrail for Safety and Adversarial Robustness in Modern LLM Systems

95 Tokenization in Transformers v5: Simpler, Clearer, and More Modular

96 The Open Evaluation Standard: Benchmarking NVIDIA Nemotron 3 Nano with NeMo Evaluator

97 CUGA on Hugging Face: Democratizing Configurable AI Agents

98 New in llama.cpp: Model Management

99 Codex is Open Sourcing AI models

100 Introducing swift-huggingface: The Complete Swift Client for Hugging Face