Unverified50% confidenceFactExact time
1-bit量化（如BitNet b1.58）将每个参数用-1、0、+1三值表示，目前主要停留在研究阶段
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
The Complete Guide to GLM-5.2: Hands-On Review of an Open-Source Long-Context Model
bilibili英文白斑马7/4/2026
Related Claims
Unverified4-bit量化只能表示16个离散值，1-bit只有2个，三元量化（-1/0/+1）只有3个可能值71% similarUnverifiedTernary quantization means each weight has only three possible value states, typically -1, 0, and +1.64% similarUnverifiedApart from bitnet-b1.58-2B-4T, there is currently no precedent for a properly trained ternary model trained from scratch.63% similarUnverified在4-bit量化下，纯参数理论最小占用为：7B约3.26GiB，13B约6.05GiB，32B约14.90GiB，70B约32.60GiB63% similarUnverifiedq4_0使用对称量化（零点固定为0），q4_1使用非对称量化（零点可浮动），后者在数值分布不对称时更精确但需额外存储零点参数62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/397154API
curl https://kongchang.com/api/v1/knowledge/claims/397154MCP
get_claim(id=397154)