Falcon-Edge：TII 推出一系列強大、通用且可微調的 1.58-bit 邊緣端語言模型★ 82
Hugging Face Blog395 days agoReleaseThe Technology Innovation Institute (TII) of the United Arab Emirates has officially released the "Falcon-Edge" model series on Hugging Face. This is a family…
微調 olmOCR 打造高保真度 OCR 引擎★ 75
Hugging Face Blog418 days agoTutorial### Background With the proliferation of vision-language models (VLMs), using VLMs for document OCR (e.g., converting PDFs to Markdown) has become mainstream…
Hugging Face 經典 NLP 課程正式轉型為 LLM 課程：迎向大語言模型時代的全面升級★ 85
Hugging Face Blog437 days agoTutorialHugging Face's "NLP Course" has long been a must-read classic for developers and researchers worldwide looking to enter the fields of Transformers and natural…
使用 Sentence Transformers 訓練與微調 Reranker 重排模型教學★ 80
Hugging Face Blog445 days agoTutorialWhen building RAG (Retrieval-Augmented Generation) systems, relying solely on vector embeddings for semantic search is often not precise enough. To improve…
Open R1 第三次更新：Hugging Face 釋出開源推理模型與 GRPO 訓練優化細節★ 85
Hugging Face Blog459 days agoReleaseSince its launch, Hugging Face's Open R1 project has been dedicated to replicating the reasoning capabilities of DeepSeek-R1 in a fully open-source manner. In…
Hugging Face 釋出 vid_ds_scripts：一站式構建影片生成高品質資料集★ 75
Hugging Face Blog487 days agoNew ToolWith the rise of open-source video generation models such as LTX-Video, HunyuanVideo, and CogVideoX, building high-quality training datasets has become the…
如何在 AWS 上部署與微調 DeepSeek 模型：Hugging Face 官方指南★ 85
Hugging Face Blog500 days agoTutorialAs DeepSeek-R1 swept through the AI landscape on the strength of its powerful reasoning capabilities, how to safely and efficiently deploy and fine-tune these…
Replicate 推出開源影片模型微調服務，支援自訂 Tencent HunyuanVideo 的風格、動作與角色★ 75
Replicate Blog506 days agoNew ToolAI cloud hosting and API services platform Replicate has announced a major update: users can now fine-tune open-source video generation models directly on the…
Hugging Face 推出 Synthetic Data Generator：用自然語言輕鬆構建 AI 訓練資料集★ 82
Hugging Face Blog545 days agoNew ToolHugging Face launched a brand-new "Synthetic Data Generator" in December 2024 — a web-based, no-code tool designed to allow anyone to create high-quality AI…
投資於效能：利用 LLM 洞察微調小型模型 — CFM 案例研究★ 75
Hugging Face Blog558 days agoBusinessThis case study from Hugging Face details how quantitative asset management firm Capital Fund Management (CFM) has optimized its investment and research…
Replicate 大幅提升 FLUX 微調速度，並將優化技術開源★ 75
Replicate Blog565 days agoReleaseThe AI cloud hosting platform Replicate has announced a major fine-tuning speed optimization for FLUX.1, currently the most popular open-source image…
Argilla 2.4 發布：在 Hugging Face Hub 上免程式碼輕鬆構建微調與評估數據集★ 75
Hugging Face Blog587 days agoReleaseThe open-source data curation and annotation platform Argilla has officially released version 2.4, with the core of this update being deep integration with…
Llama 3.2 正式支援 Keras：跨框架輕鬆微調與部署 Meta 最新輕量級模型★ 75
Hugging Face Blog601 days agoReleaseMeta's Llama 3.2 release includes lightweight 1B and 3B text models designed specifically for edge computing and mobile devices. These models have now been…
修正梯度累積：解決 LLM 微調中常被忽視的數學偏差★ 85
Hugging Face Blog606 days agoTutorial### The Mathematical Flaw in Traditional Gradient Accumulation Gradient accumulation is an extremely common technique in deep learning. When VRAM is limited…
使用合成訓練數據提升 Flux 微調效果的實用指南★ 75
Replicate Blog632 days agoTutorialFine-tuning the open-source image generation model Flux.1 has become a highly sought-after capability for creators and developers alike. However, relying…
微調 LLM 至 1.58-bit：讓極限模型量化變得簡單★ 85
Hugging Face Blog634 days agoTutorialThe deployment of large language models (LLMs) has long faced a dual bottleneck of VRAM capacity and memory bandwidth. Microsoft previously introduced the…
透過 API 微調 FLUX.1：Replicate 推出程式化微調服務★ 75
Replicate Blog643 days agoNew ToolReplicate has officially announced support for fine-tuning the popular open-source image generation model FLUX.1 [dev] via its HTTP API. FLUX.1, developed by…
微調 FLUX.1 打造專屬個人寫真：Replicate 官方 LoRA 訓練教學★ 80
Replicate Blog653 days agoTutorialThis official Replicate tutorial walks through in detail how to use LoRA (Low-Rank Adaptation) technology to fine-tune FLUX.1 [dev] — currently the most…
Replicate Intelligence #12：Flux LoRA 訓練上線、熱門的祖克柏迷因與 Lex Fridman 訪談中的 Replicate★ 75
Replicate Blog660 days agoReleaseReplicate Intelligence #12 rounds up the most noteworthy AI technical developments and community trends from late August 2024, centered on three core themes…
透過 Flash Attention 2 的 Packing 技術提升 Hugging Face 訓練效率★ 80
Hugging Face Blog662 days agoTutorialWhen fine-tuning or pre-training large language models (LLMs), the sequence lengths of input data are typically uneven. The traditional approach is to use…
Replicate Intelligence #11：微調 FLUX.1、生成式影像遊戲與元宇宙的新願景★ 75
Replicate Blog667 days agoReleaseThis edition of Replicate Intelligence #11 compiles major recent technical breakthroughs and application trends in the generative AI space, focusing primarily…
Replicate 推出 FLUX.1 微調功能：用一行程式碼即可訓練專屬的 LoRA 圖像生成模型★ 80
Replicate Blog668 days agoReleaseReplicate, the well-known cloud AI execution platform, has announced official fine-tuning support for FLUX.1, the image generation model that has taken the…
LAVE：在 Docmatix 上使用 LLM 進行零樣本 VQA 評估——我們還需要微調嗎？★ 75
Hugging Face Blog689 days agoPaper### Background and Challenges Document Visual Question Answering (DocVQA) is an important application of multimodal AI, requiring models to simultaneously…
Meta 推出 Llama 3.1：405B、70B 與 8B 旗艦開源模型，支援多語言與 128K 超長上下文★ 95
Hugging Face Blog691 days agoReleaseMeta's Llama 3.1 represents a major milestone in the open-source AI landscape. The most notable model is the 405B (405 billion parameter) version — the first…
如何利用 distilabel 打造 Argilla 2.0 專屬聊天機器人★ 75
Hugging Face Blog698 days agoTutorialIn the AI field, quickly building a chatbot that can accurately answer questions about a specific domain or newly released software has always been a major…
Replicate Intelligence #7：資料整理與資料生成的重要性
Replicate Blog702 days agoCommentaryIn the current wave of generative AI, the industry's attention is gradually shifting from "fine-tuning model architectures" to "improving data quality." Issue…
NuminaMath 如何贏得首屆 AIMO 進步獎（AI 數學奧林匹亞）並宣佈完整開源★ 80
Hugging Face Blog703 days agoRelease### Background and Achievement The AI Mathematical Olympiad (AIMO) Progress Prize aims to advance AI models capable of solving Olympiad-level mathematical…
視覺語言模型（VLM）的偏好最佳化指南：使用 TRL 進行 DPO 微調★ 75
Hugging Face Blog704 days agoTutorialAs vision-language models (VLMs) are increasingly applied to multimodal tasks, how to make these models produce outputs that better align with human…
Hugging Face 推出全新資料集搜尋與篩選功能，大幅提升數據檢索效率★ 70
Hugging Face Blog706 days agoReleaseHugging Face's official blog announced in July 2024 the launch of new "Dataset Search and Filtering Features," aimed at addressing the pain point of precisely…
微調 Microsoft Florence-2：微軟頂尖視覺語言模型實戰指南★ 80
Hugging Face Blog720 days agoTutorialMicrosoft open-sourced Florence-2 in June 2024 — a vision-language model (VLM) based on a sequence-to-sequence architecture. Despite its compact size (the Base…