Both formats store the same eight 4-bit values per element. The difference is all in the scale, and the answer shows which of the two scale choices buys more accuracy and what it costs in bits.
an account raises the free count per topic
← Quantization & Numerics / 06
mediumNewNVIDIAAMDOpenAI
NVFP4 versus MXFP4: block size, scale format and bits per element, and why does the smaller block with a finer scale help accuracy?
Both formats store the same eight 4-bit values per element. The difference is all in the scale, and the answer shows which of the two scale choices buys more accuracy and what it costs in bits.
Updated Oct 2026 · Worked answers on inference mechanisms, with calculations, implementation details, and follow-up questions.
UP NEXT ON YOUR JOURNEY
Next in this trackYou quantized a model and perplexity barely moved. How do you evaluate it before it serves traffic, and why is perplexity not enough?Next in this trackWhich layers should stay in higher precision after quantization, and how do you find them with a sensitivity sweep?Next in this trackIn a block-scaled FP8 GEMM like DeepSeek V3's, where are the scales applied, and why must accumulation be promoted to FP32?
DISCUSSION · 0
Nobody has commented on this answer. Add a correction or another way to work it.