sail-sg/Adan
Adan: Adaptive Nesterov Momentum Algorithm for Faster Optimizing Deep Models
When training deep learning models, Adan is an optimization algorithm designed to speed up the process. It takes your model's parameters and a learning rate as input, and outputs optimized parameters that help your model learn faster. This is intended for machine learning practitioners and researchers who are developing and training advanced deep learning models across various domains.
808 stars. No commits in the last 6 months.
Use this if you are training large deep learning models like Vision Transformers, BERT, GPT-2, or text-to-3D generation models and want to achieve faster convergence and potentially use higher learning rates than traditional optimizers.
Not ideal if you are working with simpler machine learning models or if memory footprint on a single GPU is a critical constraint without the option for distributed training.
Stars
808
Forks
71
Language
Python
License
Apache-2.0
Category
Last pushed
Jun 08, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/sail-sg/Adan"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
nschaetti/EchoTorch
A Python toolkit for Reservoir Computing and Echo State Network experimentation based on...
metaopt/torchopt
TorchOpt is an efficient library for differentiable optimization built upon PyTorch.
gpauloski/kfac-pytorch
Distributed K-FAC preconditioner for PyTorch
opthub-org/pytorch-bsf
PyTorch implementation of Bezier simplex fitting
pytorch/xla
Enabling PyTorch on XLA Devices (e.g. Google TPU)