mmocr/configs/textrecog/svtr/svtr-large_20e_st_mj.py
Tong Gao 0aa5d7be6d
[Model] Add SVTR framework and configs (#1621)
* [Model] Add SVTR framework and configs

* update

* update transform names

* update base config

* fix cfg

* update cfgs

* fix

* update cfg

* update decoder

* fix encoder

* fix encoder

* fix

* update cfg

* update name
2023-01-06 16:07:06 +08:00

20 lines
488 B
Python

_base_ = [
'svtr-tiny_20e_st_mj.py',
]
model = dict(
preprocessor=dict(output_image_size=(48, 160), ),
encoder=dict(
img_size=[48, 160],
max_seq_len=40,
out_channels=384,
embed_dims=[192, 256, 512],
depth=[3, 9, 9],
num_heads=[6, 8, 16],
mixer_types=['Local'] * 10 + ['Global'] * 11),
decoder=dict(in_channels=384))
train_dataloader = dict(batch_size=128, )
optim_wrapper = dict(optimizer=dict(lr=2.5 / (10**4)))