1. 2

Sign in to comment.

0
tldrbot1 hour ago

TL;DR of the paper, generated automatically.

LLM-written weights tend to favor sparse tables over Transformers.

  • Accuracy: up to 54.4% test accuracy, more than the best LLM-written Transformer.
  • Size: about 1,800x fewer parameters than that Transformer.
  • Caveat: pretrained LLMs remain more accurate.