Consilium：當多個 LLM 協同工作時的共識與協作機制★ 75
Hugging Face Blog333 days agoOpinionAs large language models (LLMs) have developed, the industry has gradually come to recognize the limitations of the "single model does everything" approach…
Ettin Suite：Hugging Face 推出最先進的成對編碼器與解碼器開源套件★ 75
Hugging Face Blog334 days agoReleaseHugging Face has officially launched the Ettin Suite, a brand-new state-of-the-art (SoTA) open-source model family of "Paired Encoders and Decoders." In…
Kimina-Prover：在大型形式化推理模型中應用測試時強化學習搜尋 (Test-time RL Search)★ 82
Hugging Face Blog339 days agoReleaseHugging Face's AI-MO (AI Math Olympiad) team has officially published Kimina-Prover, a research paper demonstrating how "test-time reinforcement learning…
Hugging Face 推出官方 MCP 伺服器：讓 AI 助理直接存取 Hub 模型與資料集★ 82
Hugging Face Blog340 days agoNew ToolHugging Face has officially announced the launch of its dedicated MCP (Model Context Protocol) server — a major step in ecosystem integration. The Model…
ScreenEnv：部署你的全端桌面 AI 代理（Desktop Agent）環境★ 82
Hugging Face Blog340 days agoNew ToolWith the rise of Anthropic's Claude 3.5 Sonnet "Computer Use" and various GUI-oriented multimodal models, "desktop agents" have become one of the hottest areas…
非同步機器人推論：解耦動作預測與執行★ 75
Hugging Face Blog340 days agoOpinionIn the fields of robot learning and embodied AI, enabling controllers based on deep learning or large language/vision models (VLAs) to run in real time has…
用 Gradio MCP 伺服器為你的大型語言模型（LLM）升級技能★ 82
Hugging Face Blog341 days agoReleaseWith Anthropic's introduction of the Model Context Protocol (MCP) open standard, the way AI agents connect to external tools and data sources has become…
為 AMD MI300 建立自訂 Kernel：利用 Triton 釋放 AMD GPU 的極致效能★ 72
Hugging Face Blog341 days agoTutorialAs AMD Instinct MI300 series GPUs (such as the MI300X) gradually increase their market share in the AI compute market, how to perform low-level optimization…
Reachy Mini：為當下與未來 AI 開發者打造的開源機器人★ 75
Hugging Face Blog341 days agoReleaseHugging Face recently announced a collaboration with Pollen Robotics to launch a new open-source robotics platform called "Reachy Mini," designed to provide…
Hugging Face 發表 SmolLM3：輕量、多語言、長上下文的端側推理模型★ 80
Hugging Face Blog342 days agoReleaseHugging Face has announced the release of a brand-new generation of lightweight open-source models — SmolLM3. As the latest member of the SmolLM family…
Hugging Face 推出高效多模態資料管線 (MMDP)：加速 VLM 與多模態模型訓練的資料處理利器★ 75
Hugging Face Blog342 days agoNew ToolWith the rapid development of vision-language models (VLMs) and multimodal AI, the amount of data required to train these models has grown explosively…
宣布 NeurIPS 2025 E2LM 競賽：聚焦語言模型的早期訓練評估
Hugging Face Blog345 days agoReleaseHugging Face and the UAE's Technology Innovation Institute (TII, the organization behind the well-known open-source model Falcon) have jointly announced a new…
使用 Sentence Transformers 訓練與微調稀疏嵌入模型 (Sparse Embedding Models)★ 75
Hugging Face Blog349 days agoTutorialThis technical blog post from Hugging Face provides a detailed guide on how to train and fine-tune "Sparse Embedding Models" using the Sentence Transformers…
Gemma 3n 正式全面登陸開源生態系！Hugging Face 完整支援指南★ 80
Hugging Face Blog354 days agoReleaseGoogle's open-source model family welcomes a new member! The all-new Gemma 3n model series is now fully available within the Hugging Face ecosystem. Gemma 3n…
SGLang 整合 Hugging Face Transformers 後端：大幅提升模型相容性與開發彈性★ 75
Hugging Face Blog357 days agoReleaseSGLang (Structured Generation Language) is a high-performance LLM inference and serving framework developed by the LMSYS team, renowned for its efficient…
在消費級硬體上微調 FLUX.1-dev：使用 QLoRA 技術指南★ 80
Hugging Face Blog361 days agoTutorialFLUX.1-dev is a state-of-the-art open-source text-to-image model with 12 billion parameters (12B), developed by Black Forest Labs. However, due to its enormous…
Groq 正式加入 Hugging Face 推理提供商（Inference Providers）支援極速開源模型推理★ 75
Hugging Face Blog364 days agoReleaseHugging Face announced a deep partnership with Groq, a chip company focused on ultra-fast AI inference, formally bringing Groq into the Hugging Face "Inference…
長 Prompt 如何阻塞其他請求？優化 LLM 推理效能與解決隊頭阻塞的關鍵策略★ 80
Hugging Face Blog367 days agoTutorialAs the context windows of large language models (LLMs) continue to expand — from the early 4k and 8k, to the now-common 32k and even 128k or more — users have…
Featherless AI 正式加入 Hugging Face 推理供應商（Inference Providers）★ 75
Hugging Face Blog368 days agoReleaseHugging Face officially announced a partnership with Featherless AI, a serverless GPU inference platform, integrating it into the Hugging Face Inference…
5 分鐘快速上手 Hugging Face Kernel Hub：GPU 加速算子託管新起點★ 78
Hugging Face Blog368 days agoReleaseThe Hugging Face official blog published a "Get Started with Hugging Face Kernel Hub in 5 Minutes" tutorial, formally introducing this new platform to the…
Hugging Face 聯手 NVIDIA 推出全新「訓練集群即服務」(Training Cluster as a Service)★ 85
Hugging Face Blog369 days agoReleaseHugging Face has announced a new partnership with AI chip giant NVIDIA, launching "Training Cluster as a Service" (TCaaS). The introduction of this service…
從零開始在 nanoVLM 中實作 KV Cache★ 75
Hugging Face Blog376 days agoTutorialIn the inference process of large language models (LLMs) and vision-language models (VLMs), autoregressive decoding is a major performance bottleneck. Each…
在 Arm 架構上實現即時 AI 聲音生成：賦予創意自由的個人工具
Hugging Face Blog376 days agoReleaseAs generative AI technology becomes more widespread, AI Sound Generation has become an indispensable part of modern multimedia creation, game development, and…
SmolVLA：基於 LeRobot 社群數據訓練的高效視覺-語言-動作（VLA）模型★ 75
Hugging Face Blog377 days agoReleaseHugging Face has recently taken an important step in the field of embodied AI, officially launching **SmolVLA** — a lightweight Vision-Language-Action (VLA)…
讓 GPU 毫無閒置：利用 TRL 中協同部署的 vLLM 解鎖高效能強化學習訓練★ 85
Hugging Face Blog377 days agoReleaseIn the reinforcement learning from human feedback (RLHF) training process for large language models — whether PPO or the recently popular GRPO — there are…
CodeAgents + Structure：更優雅、高效的 Agent 行動執行新範式★ 75
Hugging Face Blog383 days agoOpinionIn this Hugging Face blog post, the team takes a deep dive into the evolution of AI agent architectures — specifically how to combine "structured constraints"…
🐯 Liger GRPO 攜手 TRL：大幅降低 DeepSeek-R1 式強化學習訓練顯存與加速★ 82
Hugging Face Blog386 days agoNew ToolSince the explosive rise of DeepSeek-R1, GRPO (Group Relative Policy Optimization) has become the most widely discussed reinforcement learning (RL) technique…
在 Python 中打造微型 Agent：用約 70 行程式碼實現 MCP 驅動的智慧代理★ 80
Hugging Face Blog388 days agoTutorialHugging Face recently published a highly practical technical tutorial demonstrating how to build a fully functional miniature AI agent in just around 70 lines…
Dell Enterprise Hub 助企業輕鬆在本地端建置 AI 應用★ 75
Hugging Face Blog388 days agoReleaseAs enterprises place ever-increasing demands on data privacy, security, and regulatory compliance, deploying AI models on-premises has become the preferred…
Falcon-H1：重新定義效率與性能的混合頭（Hybrid-Head）語言模型系列★ 75
Hugging Face Blog390 days agoReleaseThe Technology Innovation Institute (TII) of the UAE recently officially unveiled a brand-new open-source language model series on the Hugging Face blog —…