GGUFFiles

Introduction

GGUFFiles is a Julia package for reading and writing GGUF (GGUF: GGML Unified File Format) files. It provides the necessary tools to parse, create, and manipulate GGUF files, which are used for storing and distributing machine learning models.

Features

• Reading GGUF headers, metadata, and tensor information.
• Writing GGUF headers, metadata, and tensor information.
• Support for various GGUF value types

Installation

To install GGUFFiles, use the Julia package manager:

import Pkg
Pkg.add("GGUFFiles")

Usage

Here's a basic example of how to parse a GGUF file:

using GGUFFiles

# Replace "your_gguf_file.gguf" with the path to your GGUF file
file_path = "your_gguf_file.gguf"

try
    header, metadata_kv, tensor_info = parse_gguf(file_path)

    println("Header: ", header)
    println("Metadata count: ", length(metadata_kv))
    println("Tensor count: ", length(tensor_info))
catch e
    println("Error parsing GGUF file: ", e)
end

And here's an example of how to create a GGUF file:

using GGUFFiles

# Define header
header = GGUFHeader(GGUF_MAGIC, GGUF_VERSION, 0, 0)

# Define metadata
metadata_kv = Vector{GGUFMetadataKV}()

# Define tensor info
tensor_info = Vector{GGUFTensorInfo}()

# Replace "your_gguf_file.gguf" with the desired file path
file_path = "your_gguf_file.gguf"

try
    create_gguf(file_path, header, metadata_kv, tensor_info)
    println("GGUF file created successfully.")
catch e
    println("Error creating GGUF file: ", e)
end

API Reference

Functions

• create_gguf(file_path::String, header::GGUFHeader, metadata_kv::Vector{GGUFMetadataKV}, tensor_info::Vector{GGUFTensorInfo}): Creates a GGUF file.

• parse_gguf(file_path::String): Parses a GGUF file.

• Full API is listed in the docs.