{
  "_name_or_path": "AlexHung29629/test_mllama_v17",
  "architectures": [
    "UltravoxModel"
  ],
  "audio_model_id": "AlexHung29629/wav2vec2-lv-60-espeak-cv-ft",
  "auto_map": {
    "AutoConfig": "ultravox_config.UltravoxConfig",
    "AutoModel": "ultravox_model.UltravoxModel",
    "AutoProcessor": "ultravox_processing.UltravoxProcessor"
  },
  "custom_pipelines": {
    "ultravox-pipeline": {
      "impl": "ultravox_pipeline.UltravoxPipeline",
      "pt": [
        "AutoModel"
      ],
      "tf": [],
      "type": "multimodal"
    }
  },
  "hidden_size": 4096,
  "ignore_index": -100,
  "initializer_range": 0.02,
  "model_type": "ultravox",
  "norm_init": 0.4,
  "pad_token_id": 128009,
  "projector_act": "swiglu",
  "stack_factor": 4,
  "text_model_id": "voidful/Llama-3.2-8B-Instruct",
  "torch_dtype": "bfloat16",
  "transformers_version": "4.46.2",
  "vocab_size": 128256
}