Skip to content
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.

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

examples/gguf-split: Change binary multi-byte units to decimal #7803

Merged
merged 1 commit into from
Jun 7, 2024

Conversation

christianazinn
Copy link
Contributor

Noticed while testing the Python convert-split PR #6942: examples/gguf-split at present uses binary multi-byte units (KiB, MiB, GiB).

Decimal units (KB, MB, GB) would be better from a standardization perspective. HuggingFace recommends 5GB per file and imposes a hard limit of 50GB, both decimal, so a user using any splitting script will be more likely to be working in decimal units to comply with these limits.

However, tools like Linux du and Windows File Explorer use binary units, which could potentially lead to user confusion.

This PR simply changes examples/gguf-split to use decimal units instead of binary units. SHA256 sums match with splits generated by the Python script #6942.

@ggerganov
Copy link
Owner

In #4101 we decided to stick to binary units for memory sizes. @slaren does it make sense to use decimal units in this context for specifying file sizes?

@slaren
Copy link
Collaborator

slaren commented Jun 7, 2024

Since in this context we are talking about the size of a file it could make sense to use decimal units, especially since the main point of this feature is to improve compatibility with huggingface. Personally I still would prefer the binary units simply because that's still the standard used in most places, other than hard drive sizes.

@ggerganov ggerganov merged commit c00fad7 into ggerganov:master Jun 7, 2024
64 of 66 checks passed
@christianazinn christianazinn deleted the gguf-split-base-fix branch June 7, 2024 12:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants