Unverified50% confidenceFactExact time
The maximum value of FP16 is approximately 65504; its dynamic range is smaller than that of BF16 and is prone to overflow issues.
1
Sources
50%
Confidence
Long-term
Relevance
8/7/2026
First Seen
Sources
Q8_K_XL与MXFP4命名之争：FP8≠Q8_0的量化真相
Related Claims
UnverifiedINT4量化将原本由65536个不同值（FP16）表示的权重压缩到仅16个离散级别75% similarVerifiedBF16 (Brain Floating Point 16) is a floating-point variant proposed by Google that retains the 8-bit exponent range of FP32 but reduces the mantissa to 7 bits.69% similarUnverifiedThe FP4 format has only 4 bits (1 sign bit, 2 exponent bits, 1 mantissa bit) and can precisely represent only approximately 16 discrete values.68% similarUnverifiedW4A4量化将权重和激活值均压缩至4位整数，理论上相比FP16可将显存占用降低至约1/467% similarUnverifiedCompared to FP16 original precision, FP8 quantization typically causes no more than 1–2% performance degradation on mainstream benchmarks.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/699426API
curl https://kongchang.com/api/v1/knowledge/claims/699426MCP
get_claim(id=699426)