为三值神经网络权重设计了高效二进制编码格式，体积缩小4倍且开销极低。
NativeTernary: A Self-Delimiting Binary Encoding with Unary Run-Length Hierarchy Markers for Ternary Neural Network Weights, Structured Data, and General Computing Infrastructure
- 采用自界定二进制编码+游程标记层级结构，实现精确2比特/权值存储
- 相比GGUF减少460倍边界开销（91字节 vs ~42KB），压缩率提升4倍
- 解码器仅10行代码，抗比特流损坏，适合部署在普通硬件上
BitNet b1.58（Ma et al., 2024）证明大型语言模型可在三值权重{-1, 0, +1}下运行，但缺乏原生二进制传输格式。NativeTernary填补此空白。在真实BitNet b1.58 2B4T架构（24层，约170个张量，20亿参数）上测试：该编码将三值权重压缩至精确2.000比特/权值，较GGUF Q2_K小1.31倍、较GGUF int8小4.0倍；同时将边界与帧开销降低460倍（91字节对比~42KB的GGUF张量头）。编码吞吐率47–69 MB/s，解码吞吐率35–45 MB/s，可在通用硬件上运行。解码器为10行无状态状态机，对比特流损坏具有鲁棒性。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。