configs/classification/imagenet/resnext/imagenet_resnext152-32x4d_jpg.py (9 lines of code) (raw):
_base_ = './resnext50-32x4d_b32x8_100e_jpg.py'
# model settings
model = dict(
backbone=dict(
type='ResNeXt',
depth=152,
groups=32,
base_width=4,
out_indices=[4], # 0: conv-1, x: stage-x
norm_cfg=dict(type='BN')))