config.json

{
  "activation": "silu",
  "attn_type": "mha",
  "bias": false,
  "d_model": 1024,
  "dropout": 0.2,
  "hidden_dim": 1536,
  "kv_lora_rank": null,
  "mlp": "GLU",
  "nope_head_dim": 32,
  "num_heads": 16,
  "num_kv_heads": 16,
  "num_layers": 4,
  "q_lora_rank": null,
  "rope_head_dim": null,
  "seq_len": 256,
  "v_head_dim": null,
  "vocab_size": 50257,
  "weight_tying": true
}