Page 12, in the memory calculation for a 100-trillion-parameter model:
Current text:
“182 TB = 100 trillion parameters × 16 bits per weight × 8 bits per byte”
Suggested correction:
“182 TB ≈ 100 trillion parameters × 16 bits per weight ÷ 8 bits per byte”
The published formula multiplies by 8 bits per byte, but the conversion from bits to bytes should divide by 8. The final memory estimate is approximately correct; the unit-conversion expression appears to contain a typo.
Tried to report using errata link but link isn't valid. Please let me know if u'd prefer a different way to report errata!
Loving the book, thanks for putting this together.
Page 12, in the memory calculation for a 100-trillion-parameter model:
Current text:
“182 TB = 100 trillion parameters × 16 bits per weight × 8 bits per byte”
Suggested correction:
“182 TB ≈ 100 trillion parameters × 16 bits per weight ÷ 8 bits per byte”
The published formula multiplies by 8 bits per byte, but the conversion from bits to bytes should divide by 8. The final memory estimate is approximately correct; the unit-conversion expression appears to contain a typo.
Tried to report using errata link but link isn't valid. Please let me know if u'd prefer a different way to report errata!
Loving the book, thanks for putting this together.