config.json

{
  "dim": 4096,
  "hidden_dim": 14336,
  "head_dim": 128,
  "n_layers": 32,
  "n_heads": 32,
  "n_kv_heads": 8,
  "vocab_size": 32000,
  "norm_eps": 1e-05,
  "rope_theta": 1000000,
  "max_batch_size": 32,
  "max_seq_len": 4096,
  "moe": {
    "num_experts_per_tok": 2,
    "num_experts": 8
  },
  "rope_scaling": null,
  "lora_rank": 16,
  "bias_tuning": true,
  "norm_tuning": false
}