{
  "architectures": [
    "LDT"
  ],
  "vocab_size": 12288,
  "d_model": 320,
  "n_layers": 5,
  "n_heads": 5,
  "ffn_dim": 896,
  "context_length": 512,
  "tie_word_embeddings": true,
  "activation": "swiglu",
  "norm_eps": 1e-05,
  "torch_dtype": "float32"
}