mmclassification/configs/_base_/schedules/imagenet_bs256_coslr.py

# optimizer
optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(policy='CosineAnnealing', min_lr=0)
runner = dict(type='EpochBasedRunner', max_epochs=100)
Add configs for MNIST, CIFAR10 and ImageNet 2020-07-09 23:59:10 +08:00			`# optimizer`
			`optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)`
			`optimizer_config = dict(grad_clip=None)`
			`# learning policy`
fix coslr (#32) Co-authored-by: lixiaojie <lixiaojie@sensetime.com> 2020-08-13 10:45:41 +08:00			`lr_config = dict(policy='CosineAnnealing', min_lr=0)`
Use build_runner (#54) * Use build_runner in train api * Support iter in eval_hook * Add runner section * Add test_eval_hook * Pin mmcv version in install docs * Replace max_iters with max_epochs * Set by_epoch=True as default * Remove trailing space * Replace DeprecationWarning with UserWarning * pre-commit * Fix tests 2020-10-15 15:12:50 +02:00			`runner = dict(type='EpochBasedRunner', max_epochs=100)`