From Tokens to Bytes I: Scaling Hierarchical Networks | Cartesia
cartesia.ai · saved by 2 readers
A newer generation of H-Nets learns directly from raw bytes, and matches an open-source 3B Transformer with 10x less compute and data.
A newer generation of H-Nets learns directly from raw bytes, and matches an open-source 3B Transformer with 10x less compute and data.