.`iP@ddlmZddlZGddejZdS))AnyNceZdZUdZejed<dZdZdZ dde e e fdzde e e fdzde dzde dzde de de de dede dede ffd ZfdZedejfdZxZS)UltravoxConfiga This is the configuration class to store the configuration of a [`UltravoxForConditionalGeneration`]. It is used to instantiate an Ultravox model according to the specified arguments, defining the model architecture. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: audio_config (`Union[AutoConfig, dict]`, *optional*): Custom audio config or dict. text_config (`Union[AutoConfig, dict]`, *optional*): The config object of the text backbone. audio_model_id (`str`, *optional*): The model ID of the audio backbone. text_model_id (`str`, *optional*): The model ID of the text backbone. ignore_index (`int`, *optional*, defaults to -100): The ignore index for the loss function. audio_token_index (`int`, *optional*, defaults to 32000): The audio token index to encode the audio prompt. stack_factor (`int`, *optional*, defaults to 8): Audio downsampling factor for the multimodal projector. norm_init (`float`, *optional*, defaults to 0.4): The initialization value for the layer normalization. projector_act (`str`, *optional*, defaults to `"swiglu"`): The activation function used by the multimodal projector. projector_ln_mid (`bool`, *optional*, defaults to `False`): Whether to apply layer normalization at the middle of the projector or at the end. Versions v0.4.1 and below use `False`, but v0.5 and above use `True`. wrapped_model_configultravoxz <|audio|>FN}皙?swiglur audio_config text_configaudio_model_id text_model_id ignore_indexaudio_token_index hidden_size stack_factor norm_init projector_actprojector_ln_midnum_projector_layersc  ||_||_||_||_| |_| |_| |_| |_||_|4|pi}tj | dddi||_ ||_ |;d|_ |pi}tj | dddi||_tjdi| dS)N model_typellamawhisper)rrrrrrrrr transformersCONFIG_MAPPINGgetrrrsuper__init__)selfrrrrrrrrrrrrkwargs __class__s |/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/vllm/transformers_utils/configs/ultravox.pyr#zUltravoxConfig.__init__3s )!2&("* 0$8!+  %+K(4(C g66)))))D % -  !"&D '-2L , ;  y99!!!!!D  ""6"""""c|dkr|ddlm}||d|_n |dkr|ddlm}||d|_t ||S)Nrr) get_configF)trust_remote_coder)vllm.transformers_utils.configr*rrr" __setattr__)r$keyvaluer*r&s r'r-zUltravoxConfig.__setattr__`s / ! !e&7 A A A A A A(2 5E(R(R(RD % % $ $ $): A A A A A A * 5E J J JD ww""3...r(returnc4|jS)N)rget_text_config)r$s r'rzUltravoxConfig.text_configss (88:::r() NNNNrr r r r r Fr)__name__ __module__ __qualname____doc__rPretrainedConfig__annotations__r audio_tokenis_compositiondictstrrintfloatboolr#r-propertyr __classcell__)r&s@r'rr sq!!F'7777JKN/3-1%)$( !&%!&$%+#+#38nt++##s(^d*+#d +# Tz +#  +#+#+#+#+#+#+#"+#+#+#+#+#+#Z/////&;\:;;;X;;;;;r(r)typingrrr7rrr(r'rCsf n;n;n;n;n;\2n;n;n;n;n;r(