Interesting that they're saying they have a target of replicating GPT-3 with 175B parameters, but since the RETRO paper it should (in theory?) be possible to re

Interesting that they're saying they have a target of replicating GPT-3 with 175B parameters, but since the RETRO paper it should (in theory?) be possible to replicate GPT-3 with a tenth of the parameters. Are they not planning to use a retrieval system, is it difficult to adopt their system for retrieval, or is their readme out of date?