Token Compression

topic on 1 show · 1 statements across 1 episodes

Latent Space

1 statements about Token Compression, every show

LATENT SPACE Assertion Partly supported

Scialom: Llama 3's 128K tokenizer reduces token count by about 30%

“Eight K basically, or 128 K, now with this tokenizer means 30% about less text to encode.”

Thomas Scialom Jul 23, 2024 ▶ 57:27 Training Llama 2, 3 & 4: The Path to Open Source AGI — with Thomas Scialom of Meta AI

← every entity, every show