config.json

{
  "architectures": [
    "QuadOrbitForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "configuration_quadorbit.QuadOrbitConfig",
    "AutoModelForCausalLM": "modeling_quadorbit.QuadOrbitForCausalLM"
  },
  "beta_init": 0.1,
  "beta_max": 0.5,
  "bos_token_id": 0,
  "d_ffn": 1505,
  "d_model": 512,
  "dtype": "float32",
  "dynamics_lr_scale": 0.25,
  "eos_token_id": 0,
  "eot_token_id": 0,
  "head_dim": 64,
  "init_std": 0.02,
  "max_position_embeddings": 512,
  "max_seq_len": 512,
  "model_type": "quadorbit",
  "n_heads": 8,
  "n_kv_heads": 1,
  "n_layers": 8,
  "native_gqa": false,
  "num_attention_heads": 8,
  "num_hidden_layers": 8,
  "num_key_value_heads": 1,
  "orbit_width": 8,
  "pad_token_id": 0,
  "qk_norm": true,
  "rms_eps": 1e-05,
  "rope_theta": 10000.0,
  "share_orbit_projections": true,
  "tie_embeddings": true,
  "tie_word_embeddings": true,
  "transformers_version": "5.0.0",
  "use_cache": false,
  "variant": "stable_complex_orbit_attention_swiglu",
  "vocab_size": 32768
}