config.json

{
  "vocab_size": 32768,
  "d_model": 576,
  "n_heads": 8,
  "n_kv_heads": 2,
  "n_layers": 14,
  "d_ff": 1536,
  "max_seq_len": 2048,
  "rope_theta": 500000.0,
  "rms_eps": 1e-05,
  "dropout": 0.0,
  "router_start_layer": 7,
  "gradient_checkpointing": true
}