frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-21 05:34:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BE]: Apply RUF015 to torch folder (#113025)

Browse Source 
Removes unnecessary allocations of iterators. There is a small chance this may have side effects as the entire iterator is no longer consumed, but this is a way more efficient method for retrieving the first element.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/113025
Approved by: https://github.com/ezyang, https://github.com/malfet
This commit is contained in:
Aaron Gokaslan
2023-11-07 00:48:09 +00:00
committed by PyTorch MergeBot
parent fb8ffba47f
commit 8219bf051b
26 changed files with 63 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
torch/testing/_internal/distributed/distributed_test.py
View File

@ -4875,7 +4875,7 @@ class DistributedTest:
if optimize_subset:
self.assertNotEqual(
opt_hook_init_params[0],
list(ddp_model_with_optimizer_hook.parameters())[0],
next(iter(ddp_model_with_optimizer_hook.parameters())),
)
# Untouched params should be equal
self.assertEqual(
@ -7057,7 +7057,7 @@ class DistributedTest:
# zero gradient. If we kept dividing by static initial world
# size as processes leave, the grad would be smaller.
expected_grad = torch.ones(dim, dim, device=self.rank) * grad_scale
param = list(net.parameters())[0]
param = next(iter(net.parameters()))
self.assertEqual(expected_grad, param.grad)
# Avoid accumulating grads so that it's the same every iteration
net.zero_grad()
@ -7077,7 +7077,7 @@ class DistributedTest:
* grad_scale
* effective_ws
) / dist.get_world_size()
param = list(net.parameters())[0]
param = next(iter(net.parameters()))
self.assertEqual(expected_grad, param.grad)
# Avoid accumulating grad so that it's the same every iteration.
net.zero_grad()
@ -7758,11 +7758,11 @@ class DistributedTest:
)
proxy_params = list(model.fc2.parameters())
proxy_buffers = list(model.fc2.buffers())
model_fc2_name = [
model_fc2_name = next(
module_name
for module_name, module in model.named_modules()
if module is model.fc2
][0]
)
proxy_param_names = [
f"{model_fc2_name}.{param_name}"
for param_name, _ in model.fc2.named_parameters()