config.json

{
  "architectures": [
    "UltravoxModel"
  ],
  "audio_latency_block_size": null,
  "audio_model_id": "openai/whisper-tiny",
  "auto_map": {
    "AutoConfig": "config.UltravoxConfig",
    "AutoModel": "model.UltravoxModel"
  },
  "hidden_size": 4096,
  "ignore_index": -100,
  "initializer_range": 0.02,
  "model_type": "ultravox",
  "norm_init": 0.4,
  "projector_act": "swiglu",
  "stack_factor": 8,
  "text_config": {
    "head_dim": 2,
    "hidden_size": 64,
    "model_type": "llama",
    "num_hidden_layers": 1,
    "vocab_size": 128128
  },
  "text_model_id": null,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.46.3",
  "vocab_size": 128128
}