def test_correctness_use_adaptive_pooling():
if mge.is_cuda_available():
model_name = "mnist_model_with_test.mge"
else:
model_name = "mnist_model_with_test_cpu.mge"
model_path = os.path.join(os.path.dirname(__file__), model_name)
set_execution_strategy("HEURISTIC_REPRODUCIBLE")
run_train(model_path, False, False, max_err=1e-5, use_adaptive_pooling=True)
run_train(model_path, True, False, max_err=1e-5, use_adaptive_pooling=True)
run_train(model_path, True, True, max_err=1e-5, use_adaptive_pooling=True)
# sublinear
config = SublinearMemoryConfig(genetic_nr_iter=10)
run_train(
model_path,
True,
True,
sublinear_memory_config=config,
max_err=1e-5,
use_adaptive_pooling=True,
)
run_eval(model_path, False, max_err=1e-7, use_adaptive_pooling=True)
run_eval(model_path, True, max_err=1e-7, use_adaptive_pooling=True)
def test_sublinear():
config = SublinearMemoryConfig(genetic_nr_iter=10)
@jit.trace(symbolic=True, sublinear_memory_config=config)
def f(x):
return x + x
f([0.0])
from ..builder import build_backbone
from ..registry import MODELS
def get_box(xy_ctr, offsets):
"""
xy_ctr: [1,2,37,37]
offsets: [B,2,37,37]
"""
xy0 = (xy_ctr - offsets) # top-left
xy1 = xy0 + 511 # bottom-right
bboxes_pred = F.concat([xy0, xy1], axis=1) # (B,4,H,W)
return bboxes_pred
config = SublinearMemoryConfig()
@trace(symbolic=True)
def train_generator_batch(optical, sar, label, *, opt, netG):
netG.train()
cls_score, offsets, ctr_score = netG(sar, optical)
loss, loss_cls, loss_reg, loss_ctr = netG.loss(cls_score, offsets,
ctr_score, label)
opt.backward(loss)
if dist.is_distributed():
# do all reduce mean
pass
# performance in the training data
B, _, _, _ = cls_score.shape