NativeTernary: A Self-Delimiting Binary Encoding with Unary Run-Length Hierarchy Markers for Ternary Neural Network Weights, Structured Data, and General Computing Infrastructure

arXiv:2604.03336v2 Announce Type: replace Abstract: BitNet b1.58 (Ma et al., 2024) demonstrates that large language models can operate entirely on ternary weights {-1, 0, +1}, yet no native binary wire format exists for such models. NativeTernary closes this gap. Benchmarked against GGUF on the real BitNet b1.58 2B4T architecture (24 layers, ~170 tensors, 2B parameters): NativeTernary encodes ternary weights at exactly 2.000 bits per weight -- 1.31x smaller than GGUF Q2_K and 4.0x smaller than GGUF int8 -- while reducing boundary and framing overhead by 460x (91 bytes vs ~42KB of GGUF tensor headers). Encode throughput: 47--69 MB/s. Decode throughput: 35--45 MB/s on commodity hardware. The decoder is a 10-line stateless state machine resilient to bitstream corruption.

Related

• MoBiE: Efficient Inference of Mixture of Binary Experts under Post-Training Quantization
• ShadowNPU: System and Algorithm Co-design for NPU-Centric On-Device LLM Inference
• Space Filling Curves is All You Need: Communication-Avoiding Matrix Multiplication Made Simple
• CBM-Dual: A 65-nm Fully Connected Chaotic Boltzmann Machine Processor for Dual Function Simulated Annealing and Reservoir Computing

Source: arXiv cs.LG | 2026-04-10