Unverified50% confidenceFactExact time
微软研究院2023年提出的BitNet b1.58架构将所有权重限定为{-1, 0, +1}三个状态，理论上可将模型大小压缩至原来的1/16以下
1
Sources
50%
Confidence
Long-term
Relevance
7/16/2026
First Seen
Sources
Related Claims
UnverifiedSince 2024, 1-bit quantization (such as the BitNet series) has entered academic attention, compressing weights to ternary values {-1, 0, +1}, and it is currently still in the research stage71% similarUnverified1-bit量化（BitNet）将每个权重压缩为单个比特，仅表示+1或-167% similarVerified4-bit量化理论上可将模型体积缩小8倍（相比32位存储）67% similarUnverified1-bit LLM 的核心思路是将模型权重压缩到每个参数仅约1比特的水平66% similarUnverifiedTernary representation theoretically requires only about 1.58 bits per parameter, which can yield nearly a 10x storage compression compared to FP16.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/533427API
curl https://kongchang.com/api/v1/knowledge/claims/533427MCP
get_claim(id=533427)