Yes. But standard block floating point uses a linear grid scaled by a shared exponent. Whereas AXS-6 uses a NormalFloat grid scaled by a shared exponent to maximize information density for bell-curve distributed weights. Essentially a Block Scaled Normalfloat-5.