fix np.sum overflows on windows #270
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
fascinating ty |
kw1jjang
pushed a commit
to kw1jjang/nanoGPT-BVV
that referenced
this pull request
Jun 15, 2023
fix np.sum overflows on windows
carmocca
added a commit
to Lightning-AI/litgpt
that referenced
this pull request
Jun 21, 2023
ken-viper
added a commit
to ken-viper/litgpt
that referenced
this pull request
Sep 5, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes this problem on windows: #232
Was able to reproduce on windows and fix with this.
It breaks because on windows the default value for the output of np.sum is an int32 causing overflow, while default is 64 bit on linux/mac. Same behavior with np.cumsum (the first implementation of prepare.py), so it seems like it has been overflowing from day 1 on windows, but only now people have noticed?
Tbh this just seems like stupid default behavior from numpy? Why not just force it to int64 on windows and avoid this, while letting people manually change it if needed.