• Daniël de Kok's avatar
    Make handling of FP8 scales more consisent (#2666) · 5e0fb468
    Daniël de Kok authored
    Change `fp8_quantize` so that we can pass around reciprocals everywhere,
    so scales are always passed around in the checkpoint format.
    
    I also noticed that we ignore any input scales that we might have when
    fbgemm is available. Skip this path if we already have a scale.
    5e0fb468
fp8.py 14.9 KB