config.json

{
  "attn_type": "mla",
  "dropout": 0.1,
  "hidden_dim": 768,
  "intermediate_dim": 2048,
  "kv_latent_dim": 256,
  "max_seq_len": 192,
  "n_head": 12,
  "n_kv_head": 6,
  "n_layer": 12,
  "pos_encoding": "rope",
  "rope_head_dim": 32,
  "rope_theta": 10000.0,
  "vocab_size": 1024
}