반응형
open-source package Hub (https://github.com/activeloopai/Hub).
It is a dataset format specifically for deep learning.
We have created a way to store and version-control petabyte-scale datasets as single numpy-like arrays on the cloud (and access from multiple machines simultaneously). Hub also happens to be the fastest way to access and manage datasets for PyTorch (and TensorFlow, too).
You can also use Hub as a dataloader for 120+ audio, video and image datasets (check out docs.activeloop.ai/datasets).
반응형
'AI 머신러닝' 카테고리의 다른 글
드디어 ChatGPT 4 출시를... (0) | 2023.03.14 |
---|---|
AI 활용해 벽 뒤 사물 감지하는 군사 장비 개발 (0) | 2022.07.09 |
Object Detection / Segmentation Framework, FiftyOne (0) | 2022.05.30 |
PyTorch로 시작하는 딥 러닝 입문 (0) | 2021.12.19 |
Optimizers (momentum, RMSProp, Adam, AdamW) (0) | 2021.12.07 |