You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
pytorch-image-models/timm/bits
Ross Wightman 76de984a5f
Fix some bugs with XLA support, logger, add hacky xla dist launch script since torch.dist.launch doesn't work
4 years ago
..
README.md First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
__init__.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
device_env.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
device_env_cuda.py Fix some bugs with XLA support, logger, add hacky xla dist launch script since torch.dist.launch doesn't work 4 years ago
device_env_factory.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
device_env_xla.py Fix some bugs with XLA support, logger, add hacky xla dist launch script since torch.dist.launch doesn't work 4 years ago
grad_clipper.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
logger.py Fix some bugs with XLA support, logger, add hacky xla dist launch script since torch.dist.launch doesn't work 4 years ago
tracker.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
updater.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
updater_cuda.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
updater_factory.py First timm.bits commit, add initial abstractions, WIP updates to train, val... some of it working 4 years ago
updater_xla.py Fix some bugs with XLA support, logger, add hacky xla dist launch script since torch.dist.launch doesn't work 4 years ago

README.md

Timm Bits

A collection of reusable components and lightweight abstractions for training and evaluating NN.

This is an early WIP with the primary goal to get up and running on TPUs first. Expect significant changes, rewrites, additions...

The current train.py and validate.py scipts are evolving to use the timm.bits components, they will also change significantly.