aprender-core 0.31.2

Next-generation machine learning library in pure Rust
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

//! GGUF Import/Export (spec §7.2)
//!
//! Pure Rust reader/writer for GGUF format (llama.cpp compatible).
//! WASM compatible - no C/C++ dependencies.
//!
//! # Format Structure
//!
//! ```text
//! ┌─────────────────────────────────────────┐
//! │ Magic: "GGUF" (4 bytes)                 │
//! │ Version: u32 (currently 3)              │
//! │ Tensor count: u64                       │
//! │ Metadata KV count: u64                  │
//! ├─────────────────────────────────────────┤
//! │ Metadata KV pairs                       │
//! ├─────────────────────────────────────────┤
//! │ Tensor info array                       │
//! ├─────────────────────────────────────────┤
//! │ Tensor data (aligned)                   │
//! └─────────────────────────────────────────┘
//! ```
//!
//! Reference: Gerganov, G. (2023). GGUF Format.

// Submodules (PMAT-199: split from monolithic gguf.rs)
pub mod api;
pub mod dequant;
pub mod reader;
pub mod types;

// Re-exports for backward compatibility
pub use api::*;
pub use dequant::*;
pub use reader::*;
pub use types::*;

// Module-level imports for test access via `use super::*;`
#[allow(unused_imports)]
use crate::error::{AprenderError, Result};
#[allow(unused_imports)]
use std::collections::BTreeMap;
#[allow(unused_imports)]
use std::fs::File;
#[allow(unused_imports)]
use std::io::{self, Read, Write};
#[allow(unused_imports)]
use std::path::Path;

#[cfg(test)]
mod tests;