AGCobra/gemma-gguf-header-viewer

Space

Gemma Pirate GGUF Header Viewer

0

2 commits

2 linked in READMEs

updated Sep 17, 2026

See the code
static

README

Gemma Pirate GGUF Header Viewer

A tiny browser utility that reads the 24-byte fixed header prefix of gemma-4-12b-it-pirate.Q8_0.gguf from the public AGCobra/Gemma-12b-it-Pirate model repository.

Press Read GGUF header to display the GGUF magic, format version, tensor count, metadata-entry count and raw prefix bytes. The file is pinned to revision 3887a7cd753f26397b1d95c45a84fe8e7064d58d.

The app requests bytes 0–23 with an anonymous HTTP Range request and requires the expected partial-response headers. It reads the fields as little-endian GGUF version 3, following the GGUF specification. The counts do not include the metadata values or the rest of the variable-length header.

This Space uses the model repository as a source of file-format bytes; it does not download tensor weights, perform inference or relicense the source model. The application is plain HTML and JavaScript, with no build step or server.

Contributors

AGCobra

2 commits

AGCobra/gemma-gguf-header-viewer

Space

Gemma Pirate GGUF Header Viewer

0

2 commits

2 linked in READMEs

updated Sep 17, 2026

See the code
static

README

Gemma Pirate GGUF Header Viewer

A tiny browser utility that reads the 24-byte fixed header prefix of gemma-4-12b-it-pirate.Q8_0.gguf from the public AGCobra/Gemma-12b-it-Pirate model repository.

Press Read GGUF header to display the GGUF magic, format version, tensor count, metadata-entry count and raw prefix bytes. The file is pinned to revision 3887a7cd753f26397b1d95c45a84fe8e7064d58d.

The app requests bytes 0–23 with an anonymous HTTP Range request and requires the expected partial-response headers. It reads the fields as little-endian GGUF version 3, following the GGUF specification. The counts do not include the metadata values or the rest of the variable-length header.

This Space uses the model repository as a source of file-format bytes; it does not download tensor weights, perform inference or relicense the source model. The application is plain HTML and JavaScript, with no build step or server.

Contributors

AGCobra

2 commits