Content-addressable blob storage with streaming and garbage collection
GGUF v3 parser and tensor loader for OxiLLaMa
Temporal tensor compression with tiered quantization for RuVector
Parse any tensor format, recover any precision — framework-agnostic FP8/GPTQ/AWQ/BnB dequantization, NPZ parsing, and PyTorch .pth conversion for Rust
Raw FFI bindings + dynamic loader for NVIDIA TensorRT, plus an optional C++ shim exposing a flat C ABI over TensorRT's C++-only runtime.
Sklearn-compatible adapter for ferrotorch — Tensor ↔ ndarray + curated ferrolearn re-exports
Versatile Storage Format
Unsupervised learning and clustering algorithms for ToRSh, powered by SciRS2
Compile model weight files to standalone executable binaries
WebAssembly bindings for the Tensogram N-tensor message format
Safe Rust API for NVIDIA TensorRT runtime inference.
Lib no_std Rust for native neural network (.rnn)