Hugging Face BlogFeb 12, 2025, 12:00 AMimportant 75
從 Chunks 到 Blocks：Hugging Face Hub 如何大幅加速模型與數據集的上傳與下載
Original: From Chunks to Blocks: Accelerating Uploads and Downloads on the Hub
### Background and Pain Points As large language models (LLMs) have become widespread, the file sizes hosted on the Hugging Face Hub have…
Hugging Face 官方宣布重構其 `huggingface_hub` SDK 的傳輸架構。新架構從傳統的「分塊（Chunks）」改為「區塊（Blocks）」管理，支援並行傳輸、獨立區塊重試與更低的記憶體佔用。這項更新顯著提升了數 GB 甚至數百 GB 級模型（如 Safetensors）與數據集的上傳與下載速度，並增強了網路不穩定時的容錯能力。
### Background and Pain Points
Full summary
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on Hugging Face Blog →Summaries are AI-generated; the original article is authoritative.