Reimplementation of `Improving language models by retrieving from trillions of tokens`
Primary LanguagePython
A minimal reimplementation of DeepMind's RETRO with training code.