Ma Zerun b017670e1b

[Improve] Use PyTorch official scaled_dot_product_attention to accelerate MultiheadAttention. (#1434) ...

* [Improve] Use PyTorch official `scaled_dot_product_attention` to accelerate `MultiheadAttention`.

* Support `--local-rank` and `--amp` option for new version PyTorch.

* Fix imports and UT.

2023-03-29 15:50:44 +08:00

..

apis

[Refactor] Update dev scripts to be compatible with selfsup tasks. (#1412)

2023-03-20 14:30:57 +08:00

datasets

[Docs] Update user guides docs and tools for MMPretrain. (#1429)

2023-03-27 14:32:26 +08:00

engine

[Feature] Implememnt the universal visualizer for multiple task. (#1404)

2023-03-09 11:36:54 +08:00

evaluation

[Docs] Update migration.md (#1417)

2023-03-17 10:30:09 +08:00

models

[Improve] Use PyTorch official scaled_dot_product_attention to accelerate MultiheadAttention. (#1434)

2023-03-29 15:50:44 +08:00

structures

[Refactor] Move transforms in mmselfsup to mmpretrain. (#1396)

2023-03-03 15:01:11 +08:00

utils

…

visualization

[Refactor] Refactor the browse_dataset.py to support selfsup pipeline. (#1414)

2023-03-15 14:18:36 +08:00

__init__.py

…

registry.py

…

version.py

…