[optim] abstract out _default_to_foreach_util #92305
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/92305
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 7d3259a: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
| all_tensors = [] | ||
| for tensorlist in tensorlists: | ||
| all_tensors.extend(tensorlist) | ||
| return not torch.jit.is_scripting() and not differentiable and all( |
There was a problem hiding this comment.
These booleans have nothing to do with iterating over the list, can we have it be an early return before the all_tensor list is even constructed?
There was a problem hiding this comment.
On 2nd thought, this isn't perf critical so fine as is.
There was a problem hiding this comment.
haha appreciate the perf-oriented mindset! since this is so close to landing, i'll leave it for now but will update in a later PR.
Stack from ghstack (oldest at bottom):