Module tensor_checksum 

Tensor Checksum Engine

Computes and verifies checksums for tensor data to detect corruption during storage or transmission. Supports multiple pure-Rust checksum algorithms with a stateful engine that tracks per-tensor records and verification stats.

Algorithms

Algorithm	Width	Speed	Use-case
Fnv1a64	64-bit	Fast	General-purpose, non-cryptographic hash
Adler32	32-bit	Fast	Data integrity, used in zlib
Fletcher16	16-bit	Fast	Lightweight embedded / small payloads
XorFold	64-bit	Fastest	Ultra-fast, low-collision large tensors

Examples

use ipfrs_tensorlogic::tensor_checksum::{
    ChecksumAlgorithm, TensorChecksumEngine,
};

let mut engine = TensorChecksumEngine::new();
let data = b"hello tensor world";
let record = engine.compute(1, "layer0".to_string(), data, ChecksumAlgorithm::Fnv1a64, 0);
assert!(record.is_valid(data));

let ok = engine.verify(1, data).expect("example: should succeed in docs");
assert!(ok);

Structs

ChecksumEngineStats

Aggregate statistics for a TensorChecksumEngine instance.

ChecksumRecord

Associates a TensorChecksum with a specific tensor and layer name.

TensorChecksum

A checksum value together with the metadata needed to re-verify data later.

TensorChecksumEngine

Stateful engine that manages per-tensor checksum records.

Enums

ChecksumAlgorithm

Checksum algorithm selection.

Functions

adler32

Adler-32 checksum (pure Rust).

fletcher16

Fletcher-16 checksum (pure Rust).

fnv1a64

Compute the FNV-1a 64-bit hash over arbitrary bytes.

xor_fold

XOR-fold checksum.