| #_base_ = ['./_model_base_.py',] | |
| #'https://download.openmmlab.com/mmclassification/v0/convnext/downstream/convnext-large_3rdparty_in21k_20220301-e6e0ea0a.pth' | |
| model = dict( | |
| #type='EncoderDecoderAuxi', | |
| backbone=dict( | |
| type='convnext_large', | |
| pretrained=True, | |
| in_22k=True, | |
| out_indices=[0, 1, 2, 3], | |
| drop_path_rate=0.4, | |
| layer_scale_init_value=1.0, | |
| checkpoint='data/pretrained_weight_repo/convnext/convnext_large_22k_1k_384.pth', | |
| prefix='backbones.', | |
| out_channels=[192, 384, 768, 1536]), | |
| ) | |