ZjSSKrSSKrSSKJr SSKrSSKrSSKJr SSKJ r SSK J r SSK Jr SSKJr SS KJr SS KJr SS KJr SS KJr SS KJrJrJrJrJrJr SSK J!r!J"r" SSK#J$r$ SSK%J&r&J'r'J(r( SSK)J*r* "SS\5r+"SS\RX5r-"SS\RX5r."SS\RX5r/"SS\RX5r0"SS\RX5r1SPS\RXS \RdS!\RdS"\RdS#\RdS-S$\3S-S%\3S&\$\&4S'jjr4"S(S)\RX5r5"S*S+\RX5r6"S,S-\5r7"S.S/\RX5r8"S0S1\RX5r9"S2S3\RX5r:\'"S4S5\"55r;SQS6\<\=\=4S7\3S8\=S#\R|S-S9\=S:\R~4 S;jjr@\rA\'"S<S=\;55rBS>rC\'"S?S@9"SASB\;55rD\'"SCS@9"SDSE\;55rE\'"SFSG\;55rF"SHSI\RX5rG"SJSK\RX5rH\'"SLS@9"SMSN\;55rI/SOQrJg)RN)Callable)nn)CrossEntropyLoss)initialization)ACT2FN)is_deepspeed_zero3_enabled)is_fsdp_managed_module)create_bidirectional_mask)FlashAttentionKwargs)GradientCheckpointingLayer)BaseModelOutputCausalLMOutputSequenceClassifierOutputTokenClassifierOutputWav2Vec2BaseModelOutput XVectorOutput)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstringis_peft_available)Data2VecAudioConfigc2^\rSrSrSU4SjjrSrSrU=r$)Data2VecAudioConvLayer3c>[TU]5 US:aURUS- OSUlURUUl[ R "URURURUURUURS9Ul [ R"URSS9Ul [URUlg)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 activationselfconfiglayer_id __class__s څ/root/GenerationalWealth/GenerationalWealth/venv/lib/python3.13/site-packages/transformers/models/data2vec/modeling_data2vec_audio.pyr&Data2VecAudioConvLayer.__init__4s [TU]5 US-S:XaSUlgSUlg)Nrr)r%r&num_pad_remove)r4num_conv_pos_embeddingsr7s r8r&Data2VecAudioPadLayer.__init__Os) #:Q#>!#Car:cXURS:aUSS2SS2SUR*24nU$NrrQr?s r8rAData2VecAudioPadLayer.forwardSs6    ")!Q0F43F3F2F0F*FGMr:rVrDrKs@r8rMrMNsKr:rMc.^\rSrSrU4SjrSrSrU=r$) Data2VecAudioPositionalConvLayerYcr>[TU]5 [R"URURUR UR S-UR S9Ul[UR 5Ul [URUl [R"URSS9Ulg)NrP)r paddinggroupsFr#)r%r&rr* hidden_sizeconv_pos_kernel_sizenum_conv_pos_embedding_groupsr.rMr\rr1r2r/r0r4r5r7s r8r&)Data2VecAudioPositionalConvLayer.__init__Zs II      33//1477   -V-H-HI  !?!?@,,v'9'9eTr:cURU5nURU5nURSS5nURU5nURSS5nUR U5nU$NrrP)r.r\r>r0r2r?s r8rA(Data2VecAudioPositionalConvLayer.forwardisd -0  ]3 %//15  6 %//15  6 r:)r2r.r0r\rDrKs@r8rYrYYs Ur:rYc.^\rSrSrU4SjrSrSrU=r$)$Data2VecAudioPositionalConvEmbeddingtc>[TU]5 [R"[ UR 5Vs/sHn[ U5PM sn5Ulgs snfN)r%r&r ModuleListrangerRrYlayersr4r5_r7s r8r&-Data2VecAudioPositionalConvEmbedding.__init__usF mm?DVEcEc?d e?d! -f 5?d e  esAcURSS5nURH nU"U5nM URSS5nU$rd)r>rm)r4r@layers r8rA,Data2VecAudioPositionalConvEmbedding.forward{sD%//15 [[E!-0M!%//15 r:)rmrDrKs@r8rgrgts r:rgc8^\rSrSrSrU4SjrSrSrSrU=r $)Data2VecAudioFeatureEncoderz.Construct the features from raw audio waveformc >[TU]5 [R"[ UR 5Vs/sH n[ XS9PM sn5UlSUlSUl gs snf)N)r6FT) r%r&rrkrlnum_feat_extract_layersr conv_layersgradient_checkpointing_requires_grad)r4r5ir7s r8r&$Data2VecAudioFeatureEncoder.__init__s\ ==AFvGeGeAf gAfA #F 7Af g ',#" hsA%cNUR5H nSUlM SUlgNF) parameters requires_gradr{r4params r8_freeze_parameters.Data2VecAudioFeatureEncoder._freeze_parameterss#__&E"'E '#r:cUSS2S4nUR(aUR(aSUlURH nU"U5nM U$)NT)r{trainingrry)r4 input_valuesr@ conv_layers r8rA#Data2VecAudioFeatureEncoder.forwardsK$QW-    4==*.M '**J&}5M+r:)r{ryrz) rErFrGrH__doc__r&rrArIrJrKs@r8rurus8#$  r:ruc.^\rSrSrU4SjrSrSrU=r$)Data2VecAudioFeatureProjectionc4>[TU]5 [R"URSUR S9Ul[R"URSUR5Ul [R"UR5Ul g)Nr=eps) r%r&rr/r'layer_norm_epsr0Linearr^ projectionDropoutfeat_proj_dropoutdropoutras r8r&'Data2VecAudioFeatureProjection.__init__sf ,,vr':@U@UV))FOOB$79K9KLzz&":":; r:cnURU5nURU5nURU5nX4$rj)r0rr)r4r@norm_hidden_statess r8rA&Data2VecAudioFeatureProjection.forwards7!__];(:;  ]3 00r:)rr0rrDrKs@r8rrs< 11r:rmodulequerykeyvalueattention_maskscalingrkwargsc UcURS5S-n[R"XRSS55U-nUbX-n[R R USS9n[R RXURS9n[R"X5n U RSS5R5n X4$)Nr=rPrdim)prr) sizetorchmatmulr>r functionalsoftmaxrr contiguous) rrrrrrrr attn_weights attn_outputs r8eager_attention_forwardrs**R.D(<<}}Q':;gEL!#4 ==((2(>L==((6??([L,,|3K''1-88:K  $$r:c:^\rSrSrSrSS\S\S\S\S\S \S \S-4U4S jjjr SS \ RS \ RS-S\ RS-S\S-S\ \ S\\ R\ RS-\\ RS-44 SjjrSrU=r$)Data2VecAudioAttentionz=Multi-headed attention from 'Attention Is All You Need' paperN embed_dim num_headsr is_decoderr" is_causalr5c>[TU]5 XlX lX0lX-UlXplUR U-UR:wa[SURSUS35eUR S-UlX@l X`l [R"XUS9Ul [R"XUS9Ul[R"XUS9Ul[R"XUS9Ulg)Nz;embed_dim must be divisible by num_heads (got `embed_dim`: z and `num_heads`: z).r)r")r%r&rrrhead_dimr5 ValueErrorrrrrrk_projv_projq_projout_proj) r4rrrrr"rr5r7s r8r&Data2VecAudioAttention.__init__s "" !.  MMI %$.. 8MdnnM]$YKr3 }}d* $"ii 4@ ii 4@ ii 4@  )TB r:r@key_value_statesroutput_attentionsrreturnc USLnURSSn/UQSPURP7nURU5RU5R SS5n U(aUOUn /U RSSQSPURP7n UR U 5RU 5R SS5n UR U 5RU 5R SS5n [R"URR[5nU"UU U U U4UR(dSO URURUS.UD6unnUR"/UQSP76R!5nUR#U5nUUS4$)z#Input shape: Batch x Time x ChannelNr=rrP)rrr)shaperrviewr>rrr get_interfacer5_attn_implementationrrrrreshaperr)r4r@rrrris_cross_attention input_shape hidden_shape query_statescurrent_stateskv_shape key_states value_statesattention_interfacerrs r8rAData2VecAudioAttention.forwards.T9$))#2. 88b8$--8 {{=166|DNNqRST -?)]B^))#2.BBDMMB[[055h?II!QO {{>277AKKAqQ (?(M(M KK , ,.E) %8      % $}}C$,,LL/ %  % ! \"));;;;FFH mmK0 L$..r:) r5rrrrrrrrrrr)rFTFN)NNF)rErFrGrHrintfloatboolrr&rTensorrr tuplerArIrJrKs@r8rrs G  -1CCC C  C  CC$d*CCD15.2). 0/||0/ ,,-0/ t+ 0/  $; 0/-.0/ u||U\\D0% 2E2LL M0/0/r:rc.^\rSrSrU4SjrSrSrU=r$)Data2VecAudioFeedForwardi c>[TU]5 [R"UR5Ul[R "URUR5Ul [UR[5(a[URUlOURUl[R "URUR5Ul[R"UR 5Ulgrj)r%r&rractivation_dropoutintermediate_dropoutrr^intermediate_sizeintermediate_dense isinstance hidden_actstrrintermediate_act_fn output_densehidden_dropoutoutput_dropoutras r8r&!Data2VecAudioFeedForward.__init__!s $&JJv/H/H$I!"$))F,>,>@X@X"Y f'' - -'-f.?.?'@D $'-'8'8D $IIf&>&>@R@RS jj)>)>?r:cURU5nURU5nURU5nURU5nUR U5nU$rj)rrrrrr?s r8rA Data2VecAudioFeedForward.forward.sX// > 00? 11-@ ))-8 ++M: r:)rrrrrrDrKs@r8rr s @r:rc2^\rSrSrU4SjrSSjrSrU=r$)Data2VecAudioEncoderLayeri8c>[TU]5 [URURUR SUS9Ul[R"UR5Ul [R"URURS9Ul [U5Ul[R"URURS9Ulg)NF)rrrrr5r)r%r&rr^num_attention_headsattention_dropout attentionrrrrr/rr0r feed_forwardfinal_layer_normras r8r&"Data2VecAudioEncoderLayer.__init__9s /((00,,  zz&"7"78 ,,v'9'9v?T?TU4V< " V-?-?VEZEZ [r:cUnURXUS9upnURU5nXA-nURU5nXRU5-nUR U5nU4nU(aXu4- nU$)Nrr)rrr0rr)r4r@rr attn_residualrrooutputss r8rA!Data2VecAudioEncoderLayer.forwardHs% )- L]*8* & Q ]3 %5  6 %(9(9-(HH --m<  "   &Gr:)rrrrr0rrDrKs@r8rr8s \r:rc ^\rSrSrU4SjrS S\R S\RS-S\S\S\4 S jjr S r U=r $) Data2VecAudioEncoderi\c>[TU]5 Xl[U5Ul[ R "URURS9Ul [ R"UR5Ul [ R"[UR5Vs/sHn[!U5PM sn5UlSUlgs snf)NrF)r%r&r5rgpos_conv_embedrr/r^rr0rrrrkrlnum_hidden_layersrrmrzrns r8r&Data2VecAudioEncoder.__init__]s  B6J,,v'9'9v?T?TUzz&"7"78 mmPUV\VnVnPo$pPo1%>v%FPo$pq &+#%qs C Nr@rroutput_hidden_states return_dictc,U(aSOSnU(aSOSnUb4URS5RSSURS5nSX)'[URUUS9nUR U5n XR UR5-nURU5nURU5n[5=(d [U5n URHn U(aXa4-n[R"/5n UR=(a XRR :n U (aU (a U "XUS9nUSnU (aSnU(dM|UWS4-nM U(aXa4-nU(d[#S XU455$[%UUUS 9$) Nr=rrPr)r5 inputs_embedsrrNNc3.# UH ocMUv M g7frjr).0vs r8 /Data2VecAudioEncoder.forward..sm$[q$[s )last_hidden_stater@ attentions) unsqueezerepeatrr r5rtodevicer0rr r rmrrandr layerdroprr)r4r@rrrrall_hidden_statesall_self_attentionsexpand_attention_maskposition_embeddings synced_gpusrrdropout_probabilityskip_the_layer layer_outputss r8rAData2VecAudioEncoder.forwardfs#7BD$5b4  %$2$<$(>}?S?S(TT  6  ]3 02R6LT6R [[E#$58H$H!#(**R. !]]Z/B[[EZEZ/ZN![ %!Te! !.a 0  ,   &9]1=M[TU]5 [R"URSUR-UR UR SS9Ulg)NrPr)r!r\)r%r&rr*output_hidden_sizeadapter_kernel_sizeadapter_strider.ras r8r&"Data2VecAudioAdapterLayer.__init__sJ II  % % )) )  & &((   r:cdURU5n[RRUSS9nU$)Nrr)r.rrglur?s r8rA!Data2VecAudioAdapterLayer.forwards/ -0  ))-Q)? r:)r.rDrKs@r8rrs r:rc.^\rSrSrU4SjrSrSrU=r$)Data2VecAudioAdapteric>^[TU]5 TRTR:waV[R "TRTR5Ul[R"TR5UlOS=UlUl[R"U4Sj[TR555Ul TRUl g)Nc3:># UHn[T5v M g7frj)r)rror5s r8r0Data2VecAudioAdapter.__init__..s#pOo!$=f$E$EOos)r%r&rr^rrprojr/proj_layer_normrkrlnum_adapter_layersrmrras `r8r&Data2VecAudioAdapter.__init__s   $ $(:(: : &"4"4f6O6OPDI#%<<0I0I#JD /3 3DI,mm#puU[UnUnOo#pp ))r:c|URb/URb"URU5nURU5nURSS5nURHKn[R R 5nUR (aX0R:dMCU"U5nMM URSS5nU$rd)r*r+r>rmnprandomrr)r4r@rrlayerdrop_probs r8rAData2VecAudioAdapter.forwards 99 T%9%9%E IIm4M 00?M%//15 [[EYY--/N==^nn%D %m 4 ! &//15 r:)rrmr*r+rDrKs@r8r&r&s *r:r&c\rSrSr%\\S'SrSrSrSr Sr Sr Sr \ R"5S5rSS \ R \-S \S-4S jjrSS \S \ R 4SjjrSrg)Data2VecAudioPreTrainedModelir5data2vec_audioraudioTc*[U[5(a[R"SURR - 5n[ R"URRU*US9 [ R"URRU*US9 g[U[5(a,[ R"URRS5 g[U[R5(ac[ R"URSUR R"S9 URb![ R$"UR5 gg[U[R&[R(45(a\URb [ R$"UR5 URb![ R*"UR5 gg[U[R,5(a[ R."UR5 URb_[R"UR0UR2UR4S-- 5n[ R"URU*US9 ggg)zInitialize the weightsr)abrr)meanstdN)rrmathsqrtr in_featuresinituniform_weightr"rY constant_r.rrnormal_r5initializer_rangezeros_r/ GroupNormones_r*kaiming_normal_r] in_channelsr )r4rks r8 _init_weights*Data2VecAudioPreTrainedModel._init_weightss f< = = !f//;;; MM&++00QB! <  @ A A NN6;;++Q /  * * LLSdkk6S6S T{{& FKK(' r|| < = ={{& FKK(}}( 6==))  * *   /{{&IIfmmv/A/AFDVDVWXDY/YZ[ fkkaR15'+r:N input_lengths add_adaptercdUcURROUnSn[URRURR5HupEU"XU5nM U(aD[ URR 5H!nU"USURR5nM# U$)z8 Computes the output length of the convolutional layers c8[R"X- USS9S-$)Nfloor) rounding_moder)rdiv input_lengthr r!s r8_conv_out_lengthWData2VecAudioPreTrainedModel._get_feat_extract_output_lengths.._conv_out_lengths 99\7wWZ[[ [r:r)r5rNzipr+r,rlr,r )r4rMrNrVr r!ros r8 _get_feat_extract_output_lengths=Data2VecAudioPreTrainedModel._get_feat_extract_output_lengthss 2=1Ddkk--+  \ $'t{{'>'> @W@W#X K,]PM$Y 4;;99: 04;;C]C] ^ ;r:feature_vector_lengthrcURSS9SS2S4nURXCS9nUR[R5nUR Sn[R "Xa4URURS9nSU[R"UR SURS9US- 4'URS/5RS5RS/5R5nU$)Nr=rrNr)dtyperr)r) cumsumrYr rlongrzerosr^rarangeflipr)r4r[rrNnon_padded_lengthsoutput_lengths batch_sizes r8"_get_feature_vector_attention_mask?Data2VecAudioPreTrainedModel._get_feature_vector_attention_masks ,22r2:1b5A>>?Q>k'**5::6#))!,   /~7K7KTbTiTi uv ^%9%9!%<^EZEZ[]kno]opq',,bT299"=BBB4HMMOr:rrj)rErFrGrHr__annotations__base_model_prefixmain_input_nameinput_modalitiessupports_gradient_checkpointing_supports_flash_attn_supports_sdpa_supports_flex_attnrno_gradrK LongTensorrrrYrgrIrr:r8r4r4s ($O&*#N ]]_662e>N>NQT>Tcgjncn,Y]%(:?:J:Jr:r4r mask_prob mask_length min_masksrc ^^^^^UunmTS:a [S5eTT:a[STSTS35e[RRS5R 5mUUUUU4SjnUb-UR 5R S5R5O[U5Vs/sHnTPM snn[R"UT4[S 9n /n U"T5n U S :XaU $UHn U"U 5n [RR[R"U TS- - 5U S S 9n[U5S :XaTS- nOUS n[R"U[R"X- [R S 9U-/5nU R#U5 M [R$"U 5n [R&"U SS2SS2S4X[T45n U R)X[T-5n [R"T5SSSS24n[R&"UX[T45R)X[T-5nU U-n U R+5TS- :a TS- XTS- :'[R,"XSS5 U $s snf) a2 Computes random mask spans for a given shape. Used to implement [SpecAugment: A Simple Data Augmentation Method for ASR](https://huggingface.co/papers/1904.08779). Note that this method is not optimized to run on TPU and should be run on CPU as part of the preprocessing during training. Args: shape: The shape for which to compute masks. This should be of a tuple of size 2 where the first element is the batch size and the second element is the length of the axis to span. mask_prob: The percentage of the whole axis (between 0 and 1) which will be masked. The number of independently generated mask spans of length `mask_length` is computed by `mask_prob*shape[1]/mask_length`. Note that due to overlaps, `mask_prob` is an upper bound and the actual percentage will be smaller. mask_length: size of the mask min_masks: minimum number of masked spans attention_mask: A (right-padded) attention mask which independently shortens the feature axis of each batch dimension. rz&`mask_length` has to be bigger than 0.zO`mask_length` has to be smaller than `sequence_length`, but got `mask_length`: z and `sequence_length`: `c>[TU-T- T-5n[UT5nUT-T:aTT-nUTS- - U:a[UTS- - S5nU$)z;Given input length, compute how many spans should be maskedrr)rmax)rUnum_masked_spanepsilonrtrsrusequence_lengths r8compute_num_masked_span6_compute_mask_indices..compute_num_masked_spanJsoi,6DwNOoy9 [ (? :-[TU]U5 Xl[U5Ul[ U5UlURS:dURS:aG[R"[R"UR5R55Ul[!U5UlUR$(a ['U5OSUlUR+5 gNr)r%r&r5rufeature_extractorrfeature_projectionmask_time_probmask_feature_probr Parameterrrr^r@masked_spec_embedrencoderrNr&adapter post_initras r8r&Data2VecAudioModel.__init__s   !G[G[chcmcm n /3/E/E/H/HI\I\/]M , ;; ( (1 ,#8)++77 KK;;++<< $ $)<<0DMaMainisis#t #74#@#G#GO]_#` 23M /r:rrrrrc >UbUOURRnUbUOURRnUbUOURRnUR U5nUR SS5nUbUR URSUSS9nURU5upURXUS9n URU UUUUS9n U Sn URbURU 5n U(d X4U SS-$[U UU RU RS 9$) a mask_time_indices (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. NrrPFr])rrrrrrr)r extract_featuresr@r )r5rrrrr>rgrrrrrData2VecAudioBaseModelOutputr@r ) r4rrrrrrrrr@encoder_outputss r8rAData2VecAudioModel.forwardsY 2C1N-TXT_T_TqTq$8$D $++JjJj &1%uEN+/*A*ABR*S' 00 ~1 ,, )/!5# ' (* << # LL7M!4qr7JJ J++-)77&11   r:)rr5rrrrrNNNNN)rErFrGrHrr&rr FloatTensorrrrrrrrrrArIrJrKs@r8rrs2"47;26 ,((,!,,t3,((4/ ,\/36:)-,0#'8 llT)8  t+8 !,,t3 8  $; 8 #Tk 8 D[8  - -8 8 r:rrPzu Data2VecAudio Model with a `language modeling` head on top for Connectionist Temporal Classification (CTC). ) custom_introc^\rSrSrU4SjrSr\SS\RS-S\RS-S\ S-S\ S-S \ S-S \RS-S \ \ -4S jj5r S r U=r$)Data2VecAudioForCTCi%c>[TU]U5 [U5Ul[R "UR 5UlURc[SURS35e[US5(aUR(a URO URn[R"X!R5UlUR#5 g)a2 config ([`Data2VecAudioForCTC`]): 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. NzYou are trying to instantiate z with a configuration that does not define the vocabulary size of the language model head. Please instantiate the model as follows: `Data2VecAudioForCTC.from_pretrained(..., vocab_size=vocab_size)`. or define `vocab_size` of your model's configuration.rN)r%r&rr5rr final_dropoutr vocab_sizerr7hasattrrNrr^rlm_headr)r4r5rr7s r8r&Data2VecAudioForCTC.__init__+s  08zz&"6"67    $00@AHH *1)G)GFL^L^F % %djdvdv yy!35F5FG  r:cLURRR5 grr5rrrs r8r*Data2VecAudioForCTC.freeze_feature_encoderF --@@Br:Nrrrrrlabelsrc UbUOURRnUbJUR5URR:a"[ SURR35eUR UUUUUS9nUSn UR U 5n URU 5n Sn UGbXUbUO"[R"U[RS9nURURS55R[R5n US:n U RS5nURU 5n[R R#U S[R$S9R'SS5n[R(R*R-S S 9 [R R/UUU UURR0URR2URR4S 9n SSS5 U(dU 4U[6S-nU bU 4U-$U$[9XUR:UR<S 9$!,(df  NL=f) a labels (`torch.LongTensor` of shape `(batch_size, target_length)`, *optional*): Labels for connectionist temporal classification. Note that `target_length` has to be smaller or equal to the sequence length of the output logits. Indices are selected in `[-100, 0, ..., config.vocab_size - 1]`. All labels set to `-100` are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size - 1]`. Nz$Label values must be <= vocab_size: rrrr=)rr^rF)enabled)blank reduction zero_infinitylosslogitsr@r )r5rryrrr5rrr ones_liker`rYrr  masked_selectrr log_softmaxfloat32r>backendscudnnflagsctc_loss pad_token_idctc_loss_reductionctc_zero_infinity_HIDDEN_STATES_START_POSITIONrr@r )r4rrrrrrrrr@rrrM labels_masktarget_lengthsflattened_targets log_probsoutputs r8rAData2VecAudioForCTC.forwardMs $&1%[TU]U5 [US5(aUR(a [ S5e[ U5UlURS-nUR(a2[R"[R"U5U- 5Ul [R"URUR 5Ul[R"UR UR$5UlUR)5 g)NrNzdSequence classification does not support the use of Data2VecAudio adapters (config.add_adapter=True)r)r%r&rrNrrr5ruse_weighted_layer_sumrrrr layer_weightsrr^classifier_proj_size projector num_labels classifierrr4r5 num_layersr7s r8r&/Data2VecAudioForSequenceClassification.__init__s  6= ) )f.@.@v 18--1  ( (!#ejj.Dz.Q!RD 6#5#5v7R7RS))F$?$?ARARS r:cLURRR5 grrrs r8r=Data2VecAudioForSequenceClassification.freeze_feature_encoderrr:cTURR5H nSUlM gz Calling this function will disable the gradient computation for the base model so that its parameters will not be updated during training. Only the classification head will be updated. FNr5rrrs r8freeze_base_model8Data2VecAudioForSequenceClassification.freeze_base_model% ((335E"'E 6r:Nrrrrrrrc 0UbUOURRnURR(aSOUnURUUUUUS9nURR(aiU[n [ R "U SS9n [RRURSS9n XRSSS5-RSS9n OUSn URU 5n UcU RSS9n OURU R SU5n U R#S5R%SSU R S5n S X)'U RSS9U RSS9RSS5- n UR'U 5nSnUbF[)5nU"URSURR*5URS55nU(dU4U[S-nUbU4U-$U$[-UUUR.UR0S 9$)  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]`, a `numpy.ndarray` or a `torch.Tensor`, *e.g.* via the torchcodec library (`pip install torchcodec`) or 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 [`Data2VecAudioProcessor.__call__`] for details. labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). NTrrrr=rrPrr)r5rrr5rrstackrrrrrrrr:rgrr r rrrrr@r )r4rrrrrrrrr@ norm_weights pooled_output padding_maskexpand_padding_maskrrloss_fctrs r8rA.Data2VecAudioForSequenceClassification.forwards0&1%->r1a-HHMMRSMTM#AJM}5  !)..1.5MBB=CVCVWXCY[ijL"."8"8"<"C"CAq-J]J]^_J`"a 25M. /)--!-4|7G7GA7G7N7S7STVXY7ZZM/  ')HFKKDKK,B,BCV[[QS_UDY)F)G!HHF)-)9TGf$ Ev E'!//))   r:)rr5rrr)rErFrGrHr&rrrrrrrrrArIrJrKs@r8rrs"C(/3)-,0#'&*C llT)C  t+C  $; C #Tk C D[ C  t#C  ) )C C r:rc^\rSrSrU4SjrSrSr\SS\RS-S\RS-S\RS-S \ S-S \ S-S \ S-S \ \ -4S jj5r SrU=r$)(Data2VecAudioForAudioFrameClassificationic>[TU]U5 [US5(aUR(a [ S5e[ U5UlURS-nUR(a2[R"[R"U5U- 5Ul [R"URUR 5UlUR UlUR%5 g)NrNzgAudio frame classification does not support the use of Data2VecAudio adapters (config.add_adapter=True)r)r%r&rrNrrr5rrrrrrrrr^rrrrs r8r&1Data2VecAudioForAudioFrameClassification.__init__s  6= ) )f.@.@y 18--1  ( (!#ejj.Dz.Q!RD ))F$6$68I8IJ ++ r:cLURRR5 grrrs r8r?Data2VecAudioForAudioFrameClassification.freeze_feature_encoderrr:cTURR5H nSUlM grrrs r8r:Data2VecAudioForAudioFrameClassification.freeze_base_modelrr:Nrrrrrrrc UbUOURRnURR(aSOUnURUUUUUS9nURR(aiU[n [ R "U SS9n [RRURSS9n XRSSS5-RSS9n OUSn URU 5n Sn UbZ[5n U "U RSUR5[ R "URSUR5SS95n U(dU 4U[S-nU$[#U U UR$UR&S 9$) rNTrrrr=r)axisr)r5rrr5rrrrrrrrrrrrargmaxrr@r )r4rrrrrrrrr@rrrrrs r8rA0Data2VecAudioForAudioFrameClassification.forward%sh0&1%->r1a-HHMMRSMTM#AJM/  ')HFKKDOOtuDY)F)G!HHFM$!//))   r:)rr5rrr)rErFrGrHr&rrrrrrrrrArIrJrKs@r8rrs C(/3&*)-,0#': llT):  t+:  t# :  $; : #Tk : D[:  & &: : r:rc2^\rSrSrSU4SjjrSrSrU=r$) AMSoftmaxLossicc>[TU]5 X0lX@lX l[ R "[R"X5SS9Ul [ R"5Ul g)NT)r) r%r&scalemarginrrrrrandnrArr)r4 input_dimrrrr7s r8r&AMSoftmaxLoss.__init__dsI   $ll5;;y#EUYZ '') r:cUR5n[RRURSS9n[RRUSS9n[ R "X5nX@R- n[RRX R5nUR[ R"UR5XT5-nURXr5nU$)Nrrr)flattenrr normalizerArmmrone_hotrrwhererr) r4r@rrA cos_thetapsionehotrrs r8rAAMSoftmaxLoss.forwardls!((!(< // 1/E HH]3 ++%&&v?ekk&++-HHyy( r:)rrrrrA)g>@g?rDrKs@r8rrcs*  r:rcf^\rSrSrSU4SjjrS\R S\R 4SjrSrU=r $) TDNNLayerizc>[TU]5 US:aURUS- OURUUlURUUlUR UUlURUUl[R"URUR -UR5Ul [R"5Ul g)Nrr)r%r&tdnn_dimr(r) tdnn_kernelr tdnn_dilationdilationrrkernelReLUr2r3s r8r&TDNNLayer.__init__{s [5(aSSKJn [5(a1[URW5(a[ R "S5 URSS5nURRRURURUR5RSS5n[RRXURR UR"S9nURSS5nUR%U5nU$)Nr) LoraLayerzDetected LoRA on TDNNLayer. LoRA weights won't be applied due to optimization. You should exclude TDNNLayer from LoRA's target modules.rrP)r%)rpeft.tuners.lorar*rr&warningswarnr>rArr)r r(rrconv1dr"r%r2)r4r@r*rAs r8rATDNNLayer.forwards    2   $++y11 O &//15 ##(():):D[TU]U5 [U5UlURS-nUR (a2[ R"[R"U5U- 5Ul [ R"URURS5Ul[[!UR55Vs/sHn[#X5PM nn[ R$"U5Ul[ R"URSS-UR(5Ul[ R"UR(UR(5Ul[/UR(UR05UlUR55 gs snf)Nrrr=rP)r%r&rr5rrrrrrrrr^r"rrlrr rktdnnxvector_output_dimrrrr objectiver)r4r5rr| tdnn_layersr7s r8r& Data2VecAudioForXVector.__init__s  08--1  ( (!#ejj.Dz.Q!RD 6#5#5vq7IJ5:3v;O5PQ5Py+5P QMM+. !#6??2+>+BFD]D]!^))F$=$=v?X?XY&v'@'@&BSBST RsFcLURRR5 grrrs r8r.Data2VecAudioForXVector.freeze_feature_encoderrr:cTURR5H nSUlM grrrs r8r)Data2VecAudioForXVector.freeze_base_modelrr:rMcXSnURRH nU"XS5nM U$)z/ Computes the output length of the TDNN layers cX- U-S-$)NrrrTs r8rVJData2VecAudioForXVector._get_tdnn_output_lengths.._conv_out_lengths!.69A= =r:r)r5r#)r4rMrVr s r8_get_tdnn_output_lengths0Data2VecAudioForXVector._get_tdnn_output_lengthss1  >  ;;22K,]KM3r:Nrrrrrrrc UbUOURRnURR(aSOUnURUUUUUS9nURR(aiU[n [ R "U SS9n [RRURSS9n XRSSS5-RSS9n OUSn URU 5n URH n U "U 5n M UcU RSS9n U R!SS9n OUR#URSS95nUR%U5n/n /n ['U5HPunnU R)U USU24RSS95 U R)U USU24R!SS95 MR [ R "U 5n [ R "U 5n [ R*"X/SS9nUR-U5nUR/U5nSnUbUR1UU5nU(dUU4U[S-nUbU4U-$U$[3UUUUR4UR6S9$) rNTrrrr=r)rr embeddingsr@r )r5rrr5rrrrrrrrrrr3r:r;rYr? enumeratercatrrr5rr@r )r4rrrrrrrrr@r tdnn_layer mean_features std_featuresfeat_extract_output_lengthstdnn_output_lengthsr|lengthstatistic_poolingoutput_embeddingsrrrs r8rAData2VecAudioForXVector.forwards0&1%->r1a-HHMMRSMTM#AJM}5 ))J&}5M$  !)..1.5M(,,,3L*.*O*OP^PbPbghPbPi*j '"&"?"?@["\ ML&':; 6$$]1gvg:%>%C%C%C%JK##M!WfW*$=$A$Aa$A$HI<"KK 6M ;;|4L!II}&CL 223DE!23  >>&&1D/07;X;Y3ZZF)-)9TGf$ Ev E(!//))   r:)rr5rrr5rr3r)rErFrGrHr&rrrrrrr?rrrrrrArIrJrKs@r8r1r1s &C( e6F6F6L /3)-,0#'&*P llT)P  t+P  $; P #Tk P D[ P  t#P   P P r:r1)rrrr1rr4rrU)Kr<r,collections.abcrnumpyr/rrtorch.nnrrr? activationsrintegrations.deepspeedr integrations.fsdpr masking_utilsr modeling_flash_attention_utilsr modeling_layersr modeling_outputsrrrrrrmodeling_utilsrrprocessing_utilsrutilsrrrconfiguration_data2vec_audiorrModulerMrYrgrurrrrrrrrrr&r4rrrrndarrayrrrrrrrrr r1__all__rr:r8r`s* $ %&!@76B9G&JJ=76BIIryy6 299 ")): 1RYY 1*!% II% <<% % << % LL4' % T\ %%'(%8R/RYYR/jryy0! :!HE 299E P $299>K?KKd/3 t c?ttt$$t+ t  t ZZ tn 7@ 5@ @ F!" i 6i  i X e -Ie  e P[ /K[ [ |BII. @ C :C  C L r: