config.json

{
  "architectures": [
    "UltravoxModel"
  ],
  "audio_latency_block_size": null,
  "audio_model_id": "openai/whisper-large-v3-turbo",
  "audio_model_lora_config": {
    "lora_alpha": 8,
    "r": 0,
    "target_modules": [
      "k_proj",
      "q_proj",
      "linear_k",
      "linear_q"
    ]
  },
  "auto_map": {
    "AutoConfig": "ultravox_config.UltravoxConfig",
    "AutoModel": "ultravox_model.UltravoxModel"
  },
  "hidden_size": 4096,
  "ignore_index": -100,
  "initializer_range": 0.02,
  "model_type": "ultravox",
  "norm_init": 0.4,
  "pad_token_id": 2,
  "projector_act": "swiglu",
  "stack_factor": 8,
  "text_model_id": "HuggingFaceTB/SmolLM2-1.7B-Instruct",
  "text_model_lora_config": {
    "lora_alpha": 8,
    "r": 0,
    "target_modules": [
      "k_proj",
      "q_proj",
      "linear_k",
      "linear_q"
    ]
  },
  "torch_dtype": "bfloat16",
  "transformers_version": "4.47.0",
  "vocab_size": 49152,
}