fast-reid/fastreid/modeling/backbones/regnet/regnetx/RegNetX-32GF_dds_8gpu.yaml

MODEL:
  TYPE: regnet
  NUM_CLASSES: 1000
REGNET:
  DEPTH: 23
  W0: 320
  WA: 69.86
  WM: 2.0
  GROUP_W: 168
OPTIM:
  LR_POLICY: cos
  BASE_LR: 0.2
  MAX_EPOCH: 100
  MOMENTUM: 0.9
  WEIGHT_DECAY: 5e-5
  WARMUP_ITERS: 5
TRAIN:
  DATASET: imagenet
  IM_SIZE: 224
  BATCH_SIZE: 256
TEST:
  DATASET: imagenet
  IM_SIZE: 256
  BATCH_SIZE: 200
NUM_GPUS: 8
OUT_DIR: .
support regnet backbone 2020-07-17 19:13:45 +08:00			`MODEL:`
			`TYPE: regnet`
			`NUM_CLASSES: 1000`
			`REGNET:`
add regnet config file 2020-07-17 19:32:36 +08:00			`DEPTH: 23`
			`W0: 320`
			`WA: 69.86`
			`WM: 2.0`
			`GROUP_W: 168`
support regnet backbone 2020-07-17 19:13:45 +08:00			`OPTIM:`
			`LR_POLICY: cos`
add regnet config file 2020-07-17 19:32:36 +08:00			`BASE_LR: 0.2`
support regnet backbone 2020-07-17 19:13:45 +08:00			`MAX_EPOCH: 100`
			`MOMENTUM: 0.9`
			`WEIGHT_DECAY: 5e-5`
make lr warmup by iter Summary: change warmup way by iter not by epoch, which will make it more flexible when training small epochs 2021-01-22 11:17:21 +08:00			`WARMUP_ITERS: 5`
support regnet backbone 2020-07-17 19:13:45 +08:00			`TRAIN:`
			`DATASET: imagenet`
			`IM_SIZE: 224`
add regnet config file 2020-07-17 19:32:36 +08:00			`BATCH_SIZE: 256`
support regnet backbone 2020-07-17 19:13:45 +08:00			`TEST:`
			`DATASET: imagenet`
			`IM_SIZE: 256`
add regnet config file 2020-07-17 19:32:36 +08:00			`BATCH_SIZE: 200`
support regnet backbone 2020-07-17 19:13:45 +08:00			`NUM_GPUS: 8`
			`OUT_DIR: .`