Zerfoo

Pure Go machine learning framework. Train, run, and serve ML models with go build. Zero CGo.

GPU-accelerated tensor, compute engine, and computation graph library for Go. CUDA/ROCm/OpenCL backends via purego — zero CGo required.

IEEE 754 half-precision (Float16) and BFloat16 arithmetic library for Go. Lossless round-trip conversion, configurable rounding modes, and full special-value support.

BPE tokenizer library for Go with HuggingFace tokenizer.json and GGUF tokenizer extraction support. Zero external dependencies.

ONNX-to-GGUF model converter CLI. Convert HuggingFace ONNX models to GGUF format for use with Zerfoo and llama.cpp. CGo-free, single static binary.

FP8 (E4M3FN) arithmetic library for Go. Configurable precision/range trade-offs for quantized ML inference and memory-constrained applications.

Homebrew tap for Zerfoo ML framework