config.json

{
  "activation": "silu",
  "attention_type": "k_is_v",
  "attn_type": "mla",
  "bias": false,
  "d_model": 1024,
  "dropout": 0.2,
  "head_dim": 64,
  "hidden_dim": 1536,
  "kv_lora_rank": null,
  "mlp": "glu",
  "num_heads": 21,
  "num_kv_heads": 0,
  "num_layers": 16,
  "rope_head_dim": 64,
  "seq_len": 256,
  "vocab_size": 50257,
  "weight_tying": true
}