Ë ©sgËãój—dZddlZddlZddlmZddlmZeje «Z Gd„de«Z y)zDac model configurationéNé)ÚPretrainedConfig)Úloggingc óV‡—eZdZdZdZdgd¢ddddd d d d f ˆfd „ Zedefd„«ZˆxZ S)Ú DacConfigaW This is the configuration class to store the configuration of an [`DacModel`]. It is used to instantiate a Dac model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the [descript/dac_16khz](https://huggingface.co/descript/dac_16khz) architecture. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: encoder_hidden_size (`int`, *optional*, defaults to 64): Intermediate representation dimension for the encoder. downsampling_ratios (`List[int]`, *optional*, defaults to `[2, 4, 8, 8]`): Ratios for downsampling in the encoder. These are used in reverse order for upsampling in the decoder. decoder_hidden_size (`int`, *optional*, defaults to 1536): Intermediate representation dimension for the decoder. n_codebooks (`int`, *optional*, defaults to 9): Number of codebooks in the VQVAE. codebook_size (`int`, *optional*, defaults to 1024): Number of discrete codes in each codebook. codebook_dim (`int`, *optional*, defaults to 8): Dimension of the codebook vectors. If not defined, uses `encoder_hidden_size`. quantizer_dropout (`bool`, *optional*, defaults to 0): Whether to apply dropout to the quantizer. commitment_loss_weight (float, *optional*, defaults to 0.25): Weight of the commitment loss term in the VQVAE loss function. codebook_loss_weight (float, *optional*, defaults to 1.0): Weight of the codebook loss term in the VQVAE loss function. sampling_rate (`int`, *optional*, defaults to 16000): The sampling rate at which the audio waveform should be digitalized expressed in hertz (Hz). Example: ```python >>> from transformers import DacModel, DacConfig >>> # Initializing a "descript/dac_16khz" style configuration >>> configuration = DacConfig() >>> # Initializing a model (with random weights) from the "descript/dac_16khz" style configuration >>> model = DacModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```Údacé@)ééér ié ir rgÐ?gð?i€>c  ó<•—||_||_||_|ddd…|_||_||_||_||_| |_|dt|«zz|_ ttj|««|_||_| |_t#‰ |Hdi| ¤Žy)Néÿÿÿÿr ©)Úencoder_hidden_sizeÚdownsampling_ratiosÚdecoder_hidden_sizeÚupsampling_ratiosÚ n_codebooksÚ codebook_sizeÚ codebook_dimÚquantizer_dropoutÚ sampling_rateÚlenÚ hidden_sizeÚintÚnpÚprodÚ hop_lengthÚcommitment_loss_weightÚcodebook_loss_weightÚsuperÚ__init__) Úselfrrrrrrrr r!rÚkwargsÚ __class__s €ú\/var/www/html/venv/lib/python3.12/site-packages/transformers/models/dac/configuration_dac.pyr#zDacConfig.__init__Ls¤ø€ð$7ˆÔ Ø#6ˆÔ Ø#6ˆÔ Ø!4±T°r°TÑ!:ˆÔØ&ˆÔØ*ˆÔØ(ˆÔØ!2ˆÔØ*ˆÔà.°!´sÐ;NÓ7OÑ2OÑPˆÔäœbŸg™gÐ&9Ó:Ó;ˆŒØ&<ˆÔ#Ø$8ˆÔ!ä ‰ÑÑ"˜6Ó"óÚreturncó„—tj|j«}tj|j |z «S)N)rrrÚmathÚceilr)r$rs r'Ú frame_ratezDacConfig.frame_ratels0€ä—W‘W˜T×3Ñ3Ó4ˆ Üy‰y˜×+Ñ+¨jÑ8Ó9Ð9r() Ú__name__Ú __module__Ú __qualname__Ú__doc__Ú model_typer#Úpropertyrr-Ú __classcell__)r&s@r'rrsQø„ñ+ ðZ€JðÚ(Ø ØØØØØ#Ø Øõ#ð@ð:˜Cò:óô:r(r) r1r+ÚnumpyrÚconfiguration_utilsrÚutilsrÚ get_loggerr.Úloggerrrr(r'úr:s9ðñã ãå3Ýð ˆ× Ñ ˜HÓ %€ôS:Ð õS:r(