2022-03-23 15:23:57 +08:00
|
|
|
_base_ = [
|
|
|
|
|
'../_base_/models/convmixer/convmixer-768-32.py',
|
|
|
|
|
'../_base_/datasets/imagenet_bs64_convmixer_224.py',
|
|
|
|
|
'../_base_/schedules/imagenet_bs1024_adamw_swin.py',
|
|
|
|
|
'../_base_/default_runtime.py',
|
|
|
|
|
]
|
|
|
|
|
|
2022-06-01 06:11:53 +00:00
|
|
|
# schedule setting
|
2022-06-02 09:11:09 +00:00
|
|
|
optim_wrapper = dict(
|
|
|
|
|
optimizer=dict(lr=0.01),
|
|
|
|
|
clip_grad=dict(max_norm=5.0),
|
|
|
|
|
)
|
2022-03-23 15:23:57 +08:00
|
|
|
|
2022-05-23 09:31:57 +00:00
|
|
|
train_cfg = dict(by_epoch=True, max_epochs=300)
|
2022-07-14 19:15:49 +08:00
|
|
|
|
2022-07-15 15:20:17 +08:00
|
|
|
# NOTE: `auto_scale_lr` is for automatically scaling LR
|
|
|
|
|
# based on the actual training batch size.
|
2022-07-14 19:15:49 +08:00
|
|
|
# base_batch_size = (10 GPUs) x (64 samples per GPU)
|
|
|
|
|
auto_scale_lr = dict(base_batch_size=640)
|
