You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
import pytorch_lightning takes about 2 seconds (~1 second when import torch has been executed before), which is relatively ~30x slower than import numpy, and ~15x slower than import pandas.
While ~2 second latency could be considered negligible for a computation-heavy applications, it becomes annoying, for example, when you try to build a CLI tool with autocompletion features. User wants quick response, but a simple tab completion would take at least ~2 seconds, just because of the slow import statement.
Hi @junsoo-ha-unknot
We already have some work started here: #12786
Lots of imports have already been refactored to load lazily.
Closing it because it's a duplicate. Thanks for reporting it!
Uh oh!
There was an error while loading. Please reload this page.
Bug description
import pytorch_lightning
takes about 2 seconds (~1 second whenimport torch
has been executed before), which is relatively ~30x slower thanimport numpy
, and ~15x slower thanimport pandas
.While ~2 second latency could be considered negligible for a computation-heavy applications, it becomes annoying, for example, when you try to build a CLI tool with autocompletion features. User wants quick response, but a simple tab completion would take at least ~2 seconds, just because of the slow import statement.
How to reproduce the bug
Error messages and logs
Environment
Current environment
More info
No response
cc @Borda
The text was updated successfully, but these errors were encountered: