config.json

{
  "d_k": 64,
  "d_model": 1024,
  "dropout_prob": 0.1,
  "gqa_div": 1,
  "l": 128,
  "max_len": 32768,
  "n_heads": 16,
  "n_layers": 24,
  "norm_eps": 1e-05,
  "scale": 1.0,
  "share_embeds": true,
  "start": 0.0,
  "theta1": 0,
  "theta2": 0.015384615384615385,
  "vocab_size": 32768
}