config.json

{
  "vocab_size": 50257,
  "n_layers": 8,
  "n_heads": 8,
  "n_kv_heads": 4,
  "embed_dim": 512,
  "ff_dim": 1792,
  "max_seq_len": 512,
  "dropout": 0.1,
  "activation": "swiglu",
  "attention_dropout": 0.1,
  "pos_encoding": "rotary",
  "tie_word_embeddings": true
}