Skip to content

Update fractional bias and test_geometric_mae to generate data with different rank #2660

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
vfdev-5 merged 5 commits into from
Aug 31, 2022
+24 −21
Merged

Update fractional bias and test_geometric_mae to generate data with different rank #2660

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
54daa6b
Update metrics to work in ddp config
puhuk Aug 20, 2022
5684cf3
Update test_geometric_mean_absolute_error.py
puhuk Aug 23, 2022
fe57bfc
Merge branch 'master' into fractional_bias_geometric_mae
sadra-barikbin Aug 26, 2022
8ea1fe4
Update test_geometric_mean_absolute_error.py
puhuk Aug 31, 2022
3d1c06d
Merge branch 'master' into fractional_bias_geometric_mae
vfdev-5 Aug 31, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 12 additions & 11 deletions tests/ignite/contrib/metrics/regression/test_fractional_bias.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,7 +113,6 @@ def _test_distrib_compute(device, tol=1e-5):
def _test(metric_device):
metric_device = torch.device(metric_device)
m = FractionalBias(device=metric_device)
torch.manual_seed(10 + rank)

y_pred = torch.randint(0, 10, size=(10,), device=device).float()
y = torch.randint(0, 10, size=(10,), device=device).float()
Expand All @@ -135,7 +134,8 @@ def _test(metric_device):

assert np_ans == pytest.approx(res, rel=tol)

for _ in range(3):
for i in range(3):
torch.manual_seed(10 + rank + i)
_test("cpu")
if device.type != "xla":
_test(idist.device())
Expand All @@ -144,22 +144,19 @@ def _test(metric_device):
def _test_distrib_integration(device, tol=1e-5):

rank = idist.get_rank()
torch.manual_seed(12)

def _test(n_epochs, metric_device):
metric_device = torch.device(metric_device)
n_iters = 80
s = 16
n_classes = 2
batch_size = 16

offset = n_iters * s
y_true = torch.rand(size=(offset * idist.get_world_size(),), dtype=torch.double).to(device)
y_preds = torch.rand(size=(offset * idist.get_world_size(),), dtype=torch.double).to(device)
y_true = torch.rand(size=(n_iters * batch_size,), dtype=torch.double).to(device)
y_preds = torch.rand(size=(n_iters * batch_size,), dtype=torch.double).to(device)

def update(engine, i):
return (
y_preds[i * s + rank * offset : (i + 1) * s + rank * offset],
y_true[i * s + rank * offset : (i + 1) * s + rank * offset],
y_preds[i * batch_size : (i + 1) * batch_size],
y_true[i * batch_size : (i + 1) * batch_size],
)

engine = Engine(update)
Expand All @@ -170,6 +167,9 @@ def update(engine, i):
data = list(range(n_iters))
engine.run(data=data, max_epochs=n_epochs)

y_preds = idist.all_gather(y_preds)
y_true = idist.all_gather(y_true)

assert "fb" in engine.state.metrics

res = engine.state.metrics["fb"]
Expand All @@ -189,7 +189,8 @@ def update(engine, i):
if device.type != "xla":
metric_devices.append(idist.device())
for metric_device in metric_devices:
for _ in range(2):
for i in range(2):
torch.manual_seed(12 + rank + i)
_test(n_epochs=1, metric_device=metric_device)
_test(n_epochs=2, metric_device=metric_device)

Expand Down
22 changes: 12 additions & 10 deletions tests/ignite/contrib/metrics/regression/test_geometric_mean_absolute_error.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,8 +99,9 @@ def get_test_cases():
]
return test_cases

for _ in range(5):
for i in range(5):
# check multiple random inputs as random exact occurencies are rare
torch.manual_seed(12 + i)
test_cases = get_test_cases()
for y_pred, y, batch_size in test_cases:
_test(y_pred, y, batch_size)
Expand Down Expand Up @@ -143,22 +144,19 @@ def _test(metric_device):
def _test_distrib_integration(device):

rank = idist.get_rank()
torch.manual_seed(12)

def _test(n_epochs, metric_device):
metric_device = torch.device(metric_device)
n_iters = 80
s = 16
n_classes = 2
batch_size = 16

offset = n_iters * s
y_true = torch.rand(size=(offset * idist.get_world_size(),)).to(device)
y_preds = torch.rand(size=(offset * idist.get_world_size(),)).to(device)
y_true = torch.rand(size=(n_iters * batch_size,)).to(device)
y_preds = torch.rand(size=(n_iters * batch_size,)).to(device)

def update(engine, i):
return (
y_preds[i * s + rank * offset : (i + 1) * s + rank * offset],
y_true[i * s + rank * offset : (i + 1) * s + rank * offset],
y_preds[i * batch_size : (i + 1) * batch_size],
y_true[i * batch_size : (i + 1) * batch_size],
)

engine = Engine(update)
Expand All @@ -169,6 +167,9 @@ def update(engine, i):
data = list(range(n_iters))
engine.run(data=data, max_epochs=n_epochs)

y_preds = idist.all_gather(y_preds)
y_true = idist.all_gather(y_true)

assert "gmae" in engine.state.metrics

res = engine.state.metrics["gmae"]
Expand All @@ -186,7 +187,8 @@ def update(engine, i):
if device.type != "xla":
metric_devices.append(idist.device())
for metric_device in metric_devices:
for _ in range(2):
for i in range(2):
torch.manual_seed(11 + rank + i)
_test(n_epochs=1, metric_device=metric_device)
_test(n_epochs=2, metric_device=metric_device)

Expand Down