S^(h`;ddlZddlZddlmZddlmZddlmZmZmZm Z m Z ddl m Z ddl mZmZmZdd lmZmZmZmZmZmZmZmZmZmZmZd d lmZdZd Z d Z!gdZ"dZ#dZ$GddejJZ&GddeZ'GddejJZ(GddejJZ)GddeejJZ*GddeZ+GddeZ,Gdd eZ-Gd!d"e eZ.d#Z/d$Z0e Z1ed%e/Gd&d'e.eZ2ed(e/Gd)d*e.eZ3ed+e/Gd,d-eZ4ed.e/Gd/d0eZ5ed1e/Gd2d3eZ6gd4Z7y)5N)nn)ACT2FN)CausalLMOutputSequenceClassifierOutputTokenClassifierOutputWav2Vec2BaseModelOutput XVectorOutput)PreTrainedModel)add_code_sample_docstringsadd_start_docstrings%add_start_docstrings_to_model_forward) Wav2Vec2AdapterWav2Vec2EncoderWav2Vec2FeatureEncoderWav2Vec2FeatureProjection#Wav2Vec2ForAudioFrameClassificationWav2Vec2ForCTC!Wav2Vec2ForSequenceClassificationWav2Vec2ForXVector Wav2Vec2ModelWav2Vec2PreTrainedModelWav2Vec2SamePadLayer)Data2VecAudioConfigrz!facebook/data2vec-audio-base-960h)ri$iz['MISTER QUILTER IS THE APOSTLE OF THE MIDDLE CLASSES AND WE ARE GLAD TO WELCOME HIS GOSPEL'g̼P@c&eZdZdfd ZdZxZS)Data2VecAudioConvLayerct||dkDr|j|dz nd|_|j||_t j |j|j|j||j||j|_ t j|jd|_ t|j|_y)Nrr) kernel_sizestridebiasTelementwise_affine)super__init__conv_dim in_conv_dim out_conv_dimrConv1d conv_kernel conv_stride conv_biasconv LayerNorm layer_normrfeat_extract_activation activation)selfconfiglayer_id __class__s q/var/www/skyplay_api_hub/venv/lib/python3.12/site-packages/transformers/models/data2vec/modular_data2vec_audio.pyr&zData2VecAudioConvLayer.__init__3s layers r7r?z,Data2VecAudioPositionalConvEmbedding.forwardssI%//15 [[ 1E!-0M 1%//15 r8r@rEs@r7rVrVls  r8rVceZdZdZy)Data2VecAudioFeatureEncoderc tjjtjt |j Dcgc]}t ||c}|_d|_d|_ ycc}w)N)r5FT) rModuler&rXrYnum_feat_extract_layersr conv_layersgradient_checkpointing_requires_grad)r3r4is r7r&z$Data2VecAudioFeatureEncoder.__init__|s\ ==AFvGeGeAf gA #FQ 7 g ',#" hsA4N)rArBrCr&rIr8r7r`r`{s#r8r`c eZdZy)Data2VecAudioFeatureProjectionNrHrIr8r7ririrJr8ric eZdZy)Data2VecAudioEncoderNrHrIr8r7rkrkrJr8rkc eZdZy)Data2VecAudioAdapterNrHrIr8r7rmrmrJr8rmc@eZdZdZeZdZdZdZdZ dZ dZ dZ dZ dZy ) Data2VecAudioPreTrainedModelz An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models. data2vec_audio input_valuesTct|trtjd|jj z }t jj|jj| |t jj|jj| |yt|tr5t jj|jjdyt|t jrm|jjj!d|j"j$|j%|jjj'yyt|t j(t j*frc|j$|jjj'|j&|jjj-dyyt|t j.rt jj1|j|jjtj|j2|j4|j6dzz }t jj|j| |yyy)zInitialize the weightsr)abr)meanstdNg?) isinstancerimathsqrt projection in_featuresrinituniform_weightr"rL constant_r.Lineardatanormal_r4initializer_rangezero_r/ GroupNormfill_r*kaiming_normal_rO in_channelsr )r3moduleks r7 _init_weightsz*Data2VecAudioPreTrainedModel._init_weightss f< = !f//;;; 'L($O&*#!N92===r8roa Data2VecAudio was proposed in [data2vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/pdf/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu and Michael Auli. This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving etc.). This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) sub-class. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior. Parameters: config ([`Data2VecAudioConfig`]): Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights. aJ Args: input_values (`torch.FloatTensor` of shape `(batch_size, sequence_length)`): Float values of input raw speech waveform. Values can be obtained by loading a *.flac* or *.wav* audio file into an array of type *List[float]* or a *numpy.ndarray*, *e.g.* via the soundfile library (*pip install soundfile*). To prepare the array into *input_values*, the [`AutoProcessor`] should be used for padding and conversion into a tensor of type *torch.FloatTensor*. See [`Wav2Vec2Processor.__call__`] for details. attention_mask (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing convolution and attention on padding token indices. Mask values selected in `[0, 1]`: - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**. [What are attention masks?](../glossary#attention-mask) `attention_mask` should be passed if the corresponding processor has `config.return_attention_mask == True`, which is the case for all pre-trained Data2Vec Audio models. Be aware that that even with `attention_mask`, zero-padded inputs will have slightly different outputs compared to non-padded inputs because there are more than one convolutional layer in the positional encodings. For a more detailed explanation, see [here](https://github.com/huggingface/transformers/issues/25621#issuecomment-1713759349). output_attentions (`bool`, *optional*): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more detail. output_hidden_states (`bool`, *optional*): Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for more detail. return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. zgThe bare Data2VecAudio Model transformer outputting raw hidden-states without any specific head on top.cleZdZdefdZdZdZeee e e e de fdZxZS)Data2VecAudioModelr4ctj|||_t||_t ||_|jdkDs|jdkDrEtjtj|jj|_t!||_|j$r t'|nd|_|j+y)Nru)ror&r4r`feature_extractorrifeature_projectionmask_time_probmask_feature_probr ParametertorchTensorrPr~masked_spec_embedrkencoder add_adapterrmadapter post_init)r3r4s r7r&zData2VecAudioModel.__init__s$--f5 !rs !.    >!"(:&uRYY6 0 ryy6 299 #"8"))# %>  ?  ? -=?4K-=`"$"#H 7m"%/5}%/ %/Pu"%/6%/ %/P#  /-N / /#  //R /  /#  /0 /  / r8