2023-09-18 08:31:06 +00:00
|
|
|
from colossalai.legacy.amp import AMP_TYPE
|
2022-11-11 09:08:17 +00:00
|
|
|
|
|
|
|
# hyperparameters
|
|
|
|
# BATCH_SIZE is as per GPU
|
|
|
|
# global batch size = BATCH_SIZE x data parallel size
|
|
|
|
BATCH_SIZE = 512
|
|
|
|
LEARNING_RATE = 3e-3
|
|
|
|
WEIGHT_DECAY = 0.3
|
2023-01-11 08:27:31 +00:00
|
|
|
NUM_EPOCHS = 2
|
|
|
|
WARMUP_EPOCHS = 1
|
2022-11-11 09:08:17 +00:00
|
|
|
|
|
|
|
# model config
|
2023-01-11 08:27:31 +00:00
|
|
|
NUM_CLASSES = 10
|
2022-11-11 09:08:17 +00:00
|
|
|
|
|
|
|
fp16 = dict(mode=AMP_TYPE.NAIVE)
|
|
|
|
clip_grad_norm = 1.0
|