4 citations · 4 across the 6 of their papers we have counts for
Showing 2023Show all
2 papers · 1 filter
cs.CL2023
MAP's not dead yet: Uncovering true language model modes by conditioning away degeneracy
Davis Yoshida, Kartik Goyal, Kevin Gimpel
It has been widely observed that exact or approximate MAP (mode-seeking) decoding from natural language generation (NLG) models consistently leads to degenerate outputs (Holtzman e…
cs.LG2023
NF4 Isn't Information Theoretically Optimal (and that's Good)
Davis Yoshida
This note shares some simple calculations and experiments related to absmax-based blockwise quantization, as used in Dettmers et al., 2023. Their proposed NF4 data type is said to…