使用 🤗 Transformers 與 Amazon SageMaker 進行分散式訓練：以 BART/T5 摘要生成模型為例
Hugging Face Blog1,899 days agoTutorialThis technical guide, published by Hugging Face in 2021, details how to use Amazon SageMaker's managed infrastructure and distributed training capabilities to…
深入理解 BigBird 的區塊稀疏注意力機制 (Block Sparse Attention)
Hugging Face Blog1,907 days agoTutorialTraditional Transformer models (such as BERT) are constrained by the quadratic complexity $O(N^2)$ of their self-attention mechanism, and are typically limited…
Amazon SageMaker 與 Hugging Face 宣布深度合作：簡化企業級 NLP 模型訓練與部署★ 75
Hugging Face Blog1,915 days agoBusinessHugging Face has officially announced a deep partnership with Amazon Web Services (AWS), aimed at natively integrating the Hugging Face Transformers platform…
使用 🤗 Transformers 在 Hugging Face 中微調 Wav2Vec2 進行英文語音辨識 (ASR)★ 70
Hugging Face Blog1,926 days agoTutorialThis is a landmark technical tutorial published by the Hugging Face team in 2021, detailing how to fine-tune Meta AI's Wav2Vec2 model using the Hugging Face…
Hugging Face 讀書會：長文本 Transformer 模型技術解析與演進
Hugging Face Blog1,929 days agoCommentaryIn the field of natural language processing (NLP), the core of standard Transformer models (such as BERT and GPT-2) is the self-attention mechanism. However…
建構華麗神經網路的實用指南：給開發者的幾個簡單建議
Hugging Face Blog1,941 days agoOpinionThis classic blog post from Hugging Face explores the common mistakes developers make when building complex (fancy) neural networks, and the simple principles…
使用 Hugging Face Transformers 與 Ray 實現大規模檢索增強生成 (RAG)
Hugging Face Blog1,956 days agoTutorialRetrieval-Augmented Generation (RAG) is a powerful architecture that combines a "retriever" with a "generator." It enables language models to dynamically…
Hugging Face 支援 PyTorch / XLA：在 Google Cloud TPU 上加速 Transformer 模型訓練
Hugging Face Blog1,957 days agoReleaseHugging Face has announced a deep collaboration with Google Cloud, officially adding support for PyTorch/XLA within its ecosystem. The goal is to address the…
Hugging Face Transformers 中的 TensorFlow 模型加速與 TF Serving 部署指南
Hugging Face Blog1,971 days agoTutorialWhen deploying Transformer models in production environments, latency and throughput are often the deciding factors for a project's success. Hugging Face…
透過 DeepSpeed 與 FairScale 的 ZeRO 技術，讓 Hugging Face 訓練容納更多參數且速度更快★ 80
Hugging Face Blog1,978 days agoReleaseAs the parameter scale of Transformer models (such as GPT, T5, etc.) grows exponentially, deep learning faces a severe "Memory Wall" challenge. With limited…
Hugging Face 如何為 API 客戶將 Transformer 推理速度提升 100 倍
Hugging Face Blog1,979 days agoReleaseIn this technical blog post, the Hugging Face team reveals in detail how they achieved up to 100x speedup in inference for Transformer models for customers of…
利用預訓練語言模型權重「熱啟動」Encoder-Decoder 模型
Hugging Face Blog2,049 days agoTutorialIn the field of natural language processing (NLP), sequence-to-sequence (Seq2Seq) models — such as those used for translation or summarization — typically…
將 Fairseq WMT19 翻譯系統移植至 Hugging Face Transformers
Hugging Face Blog2,055 days agoReleaseIn the field of natural language processing (NLP), machine translation has always been a core challenge. Facebook AI Research (FAIR) achieved outstanding…
使用 Transformers 與 Ray Tune 進行超參數搜尋
Hugging Face Blog2,056 days agoTutorialThis classic article from the official Hugging Face blog provides a detailed guide on how to integrate Hugging Face's `Transformers` library with the powerful…
Transformer 架構下的編碼器-解碼器（Encoder-Decoder）模型深度解析★ 70
Hugging Face Blog2,079 days agoTutorialThis classic blog post written by Hugging Face researcher Patrick von Platen takes a deep dive into the Transformer-based Encoder-Decoder model architecture…
使用區塊稀疏矩陣（Block Sparse Matrices）打造更小、更快的語言模型
Hugging Face Blog2,109 days agoTutorialIn the field of natural language processing (NLP), the Transformer architecture has become the dominant paradigm, but its core self-attention mechanism…
Reformer：挑戰語言模型長文本處理極限的架構
Hugging Face Blog2,178 days agoPaperThis technical blog post published by Hugging Face takes a deep dive into how the Reformer architecture overcomes the memory and computational bottlenecks that…
如何生成文本：在 Transformers 中使用不同的解碼方法進行語言生成★ 85
Hugging Face Blog2,302 days agoTutorialThis classic technical blog post written by Hugging Face takes an in-depth look at how to select and tune different "decoding methods" when performing…
如何使用 Transformers 和 Tokenizers 從頭開始訓練新的語言模型★ 75
Hugging Face Blog2,318 days agoTutorialThis classic blog post from Hugging Face provides a detailed walkthrough of how to use their open-source ecosystem libraries — `transformers` and `tokenizers`…