Fix optimizer_wrapper in Config

Yuanyang-Zhu · Jul 19, 2022 · 93817d1 · 93817d1
1 parent 127af4f
commit 93817d1
Show file tree

Hide file tree

Showing 21 changed files with 48 additions and 25 deletions.
diff --git a/configs/_base_/schedules/schedule_1x.py b/configs/_base_/schedules/schedule_1x.py
@@ -17,4 +17,6 @@
 ]
 
 # optimizer
-optimizer = dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    type='OptimWrapper',
+    optimizer=dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/_base_/schedules/schedule_20e.py b/configs/_base_/schedules/schedule_20e.py
@@ -17,4 +17,6 @@
 ]
 
 # optimizer
-optimizer = dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    type='OptimWrapper',
+    optimizer=dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/_base_/schedules/schedule_2x.py b/configs/_base_/schedules/schedule_2x.py
@@ -17,4 +17,6 @@
 ]
 
 # optimizer
-optimizer = dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    type='OptimWrapper',
+    optimizer=dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/atss/atss_r50_fpn_1x_coco.py b/configs/atss/atss_r50_fpn_1x_coco.py
@@ -67,4 +67,5 @@
         nms=dict(type='nms', iou_threshold=0.6),
         max_per_img=100))
 # optimizer
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/fcos/fcos_r101_caffe_fpn_gn-head_mstrain_640-800_2x_coco.py b/configs/fcos/fcos_r101_caffe_fpn_gn-head_mstrain_640-800_2x_coco.py
@@ -22,7 +22,7 @@
 train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
 
 # training schedule for 2x
-train_cfg = dict(by_epoch=True, max_epochs=24)
+train_cfg = dict(max_epochs=24)
 
 # learning rate
 param_scheduler = [

diff --git a/configs/fcos/fcos_r50_caffe_fpn_gn-head_1x_coco.py b/configs/fcos/fcos_r50_caffe_fpn_gn-head_1x_coco.py
@@ -74,8 +74,9 @@
 ]
 
 # optimizer
-optimizer = dict(
-    lr=0.01, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.))
+optim_wrapper = dict(
+    optimizer=dict(
+        lr=0.01, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.)))
 
 default_hooks = dict(
     optimizer=dict(

diff --git a/configs/fcos/fcos_x101_64x4d_fpn_gn-head_mstrain_640-800_2x_coco.py b/configs/fcos/fcos_x101_64x4d_fpn_gn-head_mstrain_640-800_2x_coco.py
@@ -36,7 +36,7 @@
 train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
 
 # training schedule for 2x
-train_cfg = dict(by_epoch=True, max_epochs=24)
+train_cfg = dict(max_epochs=24)
 
 # learning rate
 param_scheduler = [

diff --git a/configs/fsaf/fsaf_r50_fpn_1x_coco.py b/configs/fsaf/fsaf_r50_fpn_1x_coco.py
@@ -43,7 +43,10 @@
         allowed_border=-1,
         pos_weight=-1,
         debug=False))
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
+
 default_hooks = dict(
     optimizer=dict(
         _delete_=True,

diff --git a/configs/nas_fcos/nas_fcos_fcoshead_r50_caffe_fpn_gn-head_4x4_1x_coco.py b/configs/nas_fcos/nas_fcos_fcoshead_r50_caffe_fpn_gn-head_4x4_1x_coco.py
@@ -70,5 +70,6 @@
 train_dataloader = dict(batch_size=4, num_workers=2)
 
 # optimizer
-optimizer = dict(
-    lr=0.01, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.))
+optim_wrapper = dict(
+    optimizer=dict(
+        lr=0.01, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.)))
diff --git a/configs/nas_fcos/nas_fcos_nashead_r50_caffe_fpn_gn-head_4x4_1x_coco.py b/configs/nas_fcos/nas_fcos_nashead_r50_caffe_fpn_gn-head_4x4_1x_coco.py
@@ -69,5 +69,6 @@
 train_dataloader = dict(batch_size=4, num_workers=2)
 
 # optimizer
-optimizer = dict(
-    lr=0.01, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.))
+optim_wrapper = dict(
+    optimizer=dict(
+        lr=0.01, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.)))
diff --git a/configs/paa/paa_r50_fpn_1x_coco.py b/configs/paa/paa_r50_fpn_1x_coco.py
@@ -67,4 +67,6 @@
         nms=dict(type='nms', iou_threshold=0.6),
         max_per_img=100))
 # optimizer
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    type='OptimWrapper',
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/retinanet/retinanet_r101_fpn_mstrain_640-800_3x_coco.py b/configs/retinanet/retinanet_r101_fpn_mstrain_640-800_3x_coco.py
@@ -3,4 +3,5 @@
 ]
 # optimizer
 model = dict(pretrained='torchvision://resnet101', backbone=dict(depth=101))
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/retinanet/retinanet_r18_fpn_1x8_1x_coco.py b/configs/retinanet/retinanet_r18_fpn_1x8_1x_coco.py
@@ -15,8 +15,8 @@
     neck=dict(in_channels=[64, 128, 256, 512]))
 
 # Note: If the learning rate is set to 0.0025, the mAP will be 32.4.
-optimizer = dict(type='SGD', lr=0.005, momentum=0.9, weight_decay=0.0001)
-
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.005, momentum=0.9, weight_decay=0.0001))
 # TODO: support auto scaling lr
 # NOTE: `auto_scale_lr` is for automatically scaling LR,
 # USER SHOULD NOT CHANGE ITS VALUES.

diff --git a/configs/retinanet/retinanet_r18_fpn_1x_coco.py b/configs/retinanet/retinanet_r18_fpn_1x_coco.py
@@ -10,7 +10,8 @@
         depth=18,
         init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet18')),
     neck=dict(in_channels=[64, 128, 256, 512]))
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
 
 # TODO: support auto scaling lr
 # NOTE: `auto_scale_lr` is for automatically scaling LR,

diff --git a/configs/retinanet/retinanet_r50_fpn_1x_coco.py b/configs/retinanet/retinanet_r50_fpn_1x_coco.py
@@ -5,4 +5,5 @@
 ]
 
 # optimizer
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/retinanet/retinanet_r50_fpn_2x_coco.py b/configs/retinanet/retinanet_r50_fpn_2x_coco.py
@@ -21,4 +21,5 @@
 ]
 
 # optimizer
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/retinanet/retinanet_r50_fpn_mstrain_640-800_3x_coco.py b/configs/retinanet/retinanet_r50_fpn_mstrain_640-800_3x_coco.py
@@ -2,4 +2,5 @@
     '../_base_/models/retinanet_r50_fpn.py', '../common/mstrain_3x_coco.py'
 ]
 # optimizer
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))
diff --git a/configs/retinanet/retinanet_x101_64x4d_fpn_mstrain_640-800_3x_coco.py b/configs/retinanet/retinanet_x101_64x4d_fpn_mstrain_640-800_3x_coco.py
@@ -5,4 +5,4 @@
 model = dict(
     pretrained='open-mmlab://resnext101_64x4d',
     backbone=dict(type='ResNeXt', depth=101, groups=64, base_width=4))
-optimizer = dict(type='SGD', lr=0.01)
+optim_wrapper = dict(optimizer=dict(type='SGD', lr=0.01))
diff --git a/configs/ssd/ssd300_coco.py b/configs/ssd/ssd300_coco.py
@@ -65,7 +65,8 @@
 test_dataloader = val_dataloader
 
 # optimizer
-optimizer = dict(type='SGD', lr=2e-3, momentum=0.9, weight_decay=5e-4)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=2e-3, momentum=0.9, weight_decay=5e-4))
 
 custom_hooks = [
     dict(type='NumClassCheckHook'),

diff --git a/configs/ssd/ssdlite_mobilenetv2_scratch_600e_coco.py b/configs/ssd/ssdlite_mobilenetv2_scratch_600e_coco.py
@@ -149,7 +149,8 @@
 ]
 
 # optimizer
-optimizer = dict(type='SGD', lr=0.015, momentum=0.9, weight_decay=4.0e-5)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.015, momentum=0.9, weight_decay=4.0e-5))
 
 custom_hooks = [
     dict(type='NumClassCheckHook'),

diff --git a/configs/tood/tood_r50_fpn_1x_coco.py b/configs/tood/tood_r50_fpn_1x_coco.py
@@ -76,4 +76,5 @@
         nms=dict(type='nms', iou_threshold=0.6),
         max_per_img=100))
 # optimizer
-optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
+optim_wrapper = dict(
+    optimizer=dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001))