§
    9²¸jŒx  ã                   óV  — d dl Z d dlmZ d dlZd dlZd dlmZ ddlmZ	 ddl
mZ ddlmZmZmZ ddlmZ dd	lmZmZ dd
lmZ ddlmZmZmZmZ ddlmZmZ ddlm Z  ddl!m"Z"m#Z# ddl$m%Z%m&Z& ddl'm(Z(m)Z) ddl*m+Z+ ddl,m-Z-m.Z.  G d„ dej/        ¦  «        Z0e# G d„ de¦  «        ¦   «         Z1 G d„ dej/        ¦  «        Z2dej3        de4dej3        fd„Z5	 d=d ej/        d!ej3        d"ej3        d#ej3        d$ej3        dz  d%e6d&e6d'e e"         fd(„Z7 G d)„ d*ej/        ¦  «        Z8 G d+„ d,ej/        ¦  «        Z9 G d-„ d.e¦  «        Z:e# G d/„ d0e1¦  «        ¦   «         Z; e#d1¬2¦  «         G d3„ d4e1¦  «        ¦   «         Z<d5ej3        d6e4d7e4fd8„Z= e#d9¬2¦  «         G d:„ d;e1e¦  «        ¦   «         Z>g d<¢Z?dS )>é    N)ÚCallable)Únné   )Úinitialization)ÚACT2FN)ÚCacheÚDynamicCacheÚEncoderDecoderCache)ÚGenerationMixin)Úcreate_bidirectional_maskÚcreate_causal_mask)ÚGradientCheckpointingLayer)ÚBaseModelOutputÚ)BaseModelOutputWithPastAndCrossAttentionsÚSeq2SeqLMOutputÚSeq2SeqModelOutput)ÚALL_ATTENTION_FUNCTIONSÚPreTrainedModel)ÚUnpack)ÚTransformersKwargsÚauto_docstring)Úcan_return_tupleÚmerge_with_config_defaults)ÚOutputRecorderÚcapture_outputsé   )Ú	AutoModelé   )ÚCanaryConfigÚCanaryDecoderConfigc                   óf   ‡ — e Zd ZdZdedefˆ fd„Zdej        fd„Zdej        dej        fd„Z	ˆ xZ
S )	ÚCanaryPositionalEmbeddingzÆ
    Identical to [`SinusoidsPositionEmbedding`] except that the timescales and the `1 / sqrt(channels)` scaling match
    NeMo's `FixedPositionalEncoding`, and it is indexed by `position_ids`.
    ÚlengthÚchannelsc                 ó  •— t          ¦   «                              ¦   «          || _        || _        d|dz
  |z  z  | _        |dz  dk    rt          d¦  «        ‚|                      ¦   «         }t          j        |d¬¦  «        | _	        d S )Ni'  r   r   z3CanaryPositionalEmbedding needs even channels inputF)Ú
persistent)
ÚsuperÚ__init__r#   r$   Úmax_timescaleÚ
ValueErrorÚ-compute_default_singular_positional_embeddingr   ÚBufferÚpositional_embedding)Úselfr#   r$   Úposition_embeddingÚ	__class__s       €úz/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/transformers/models/canary/modeling_canary.pyr(   z"CanaryPositionalEmbedding.__init__7   s‹   ø€ Ý‰Œ×ÒÑÔÐØˆŒØ ˆŒØ"¨°1©¸Ñ'@ÑAˆÔØ�a‰<˜1ÒÐÝÐRÑSÔSÐSØ!×OÒOÑQÔQÐÝ$&¤IÐ.@ÈUÐ$SÑ$SÔ$SˆÔ!Ð!Ð!ó    Úreturnc                 óØ  — t          j        | j        ¦  «        | j        dz  dz
  z  }t	          j        | t	          j        | j        dz  ¦  «                             ¦   «         z  ¦  «        }t	          j        | j        ¦  «        d d …t           j	        f         |t           j	        d d …f         z  }t	          j
        t	          j        |¦  «        t	          j        |¦  «        gd¬¦  «        S )Nr   r   )Údim)ÚnpÚlogr)   r$   ÚtorchÚexpÚarangeÚfloatr#   ÚnewaxisÚcatÚsinÚcos)r.   Úlog_timescale_incrementÚinv_timescalesÚscaled_times       r1   r+   zGCanaryPositionalEmbedding.compute_default_singular_positional_embeddingA   sÆ   € Ý"$¤&¨Ô);Ñ"<Ô"<ÀÄÐQRÑ@RÐUVÑ@VÑ"WÐÝœÐ$;Ð#;½e¼lÈ4Ì=Ð\]ÑK]Ñ>^Ô>^×>dÒ>dÑ>fÔ>fÑ#fÑgÔgˆÝ”l 4¤;Ñ/Ô/°°°µ2´:°Ô>ÀÕPRÔPZÐ\]Ð\]Ð\]ÐP]ÔA^Ñ^ˆÝŒy�%œ) KÑ0Ô0µ%´)¸KÑ2HÔ2HÐIÈqÐQÑQÔQÐQr2   Úposition_idsc                 óP   — | j         |         t          j        | j        ¦  «        z  S ©N)r-   ÚmathÚsqrtr$   )r.   rC   s     r1   Úforwardz!CanaryPositionalEmbedding.forwardG   s!   € ØÔ(¨Ô6½¼À4Ä=Ñ9QÔ9QÑQÐQr2   )Ú__name__Ú
__module__Ú__qualname__Ú__doc__Úintr(   r8   ÚTensorr+   rH   Ú__classcell__©r0   s   @r1   r"   r"   1   s¯   ø€ € € € € ðð ð
T˜sð T¨cð Tð Tð Tð Tð Tð TðR¸u¼|ð Rð Rð Rð RðR E¤Lð R°U´\ð Rð Rð Rð Rð Rð Rð Rð Rr2   r"   c                   ó|   ‡ — e Zd ZU eed<   dZdZdZdZdgZ	dZ
dZdZdgZ ej        ¦   «         ˆ fd„¦   «         Zˆ xZS )	ÚCanaryPreTrainedModelÚconfigÚmodelÚinput_featuresÚaudioTÚCanaryDecoderLayerzpreprocessor\.featurizer\..*c                 óÐ   •— t          ¦   «                              |¦  «         t          |t          ¦  «        r.t	          j        |j        |                     ¦   «         ¦  «         d S d S rE   )r'   Ú_init_weightsÚ
isinstancer"   ÚinitÚcopy_r-   r+   )r.   Úmoduler0   s     €r1   rY   z#CanaryPreTrainedModel._init_weightsY   sb   ø€ å‰Œ×Ò˜fÑ%Ô%Ð%Ý�fÕ7Ñ8Ô8ð 	lÝŒJ�vÔ2°F×4hÒ4hÑ4jÔ4jÑkÔkÐkÐkÐkð	lð 	lr2   )rI   rJ   rK   r   Ú__annotations__Úbase_model_prefixÚmain_input_nameÚinput_modalitiesÚsupports_gradient_checkpointingÚ_no_split_modulesÚ_supports_flash_attnÚ_supports_sdpaÚ_can_compile_fullgraphÚ"_keys_to_ignore_on_load_unexpectedr8   Úno_gradrY   rO   rP   s   @r1   rR   rR   K   s›   ø€ € € € € € àÐÐÑØÐØ&€OØÐØ&*Ð#Ø-Ð.ÐØÐØ€Nà!ÐØ*IÐ)JÐ&à€U„]�_„_ðlð lð lð lñ „_ðlð lð lð lð lr2   rR   c                   óB   ‡ — e Zd Zˆ fd„Zdej        dej        fd„Zˆ xZS )ÚCanaryDecoderMLPc                 ó  •— t          ¦   «                              ¦   «          || _        t          |j                 | _        t          j        |j        |j	        ¦  «        | _
        t          j        |j	        |j        ¦  «        | _        d S rE   )r'   r(   rS   r   Ú
hidden_actÚactivation_fnr   ÚLinearÚhidden_sizeÚintermediate_sizeÚfc1Úfc2©r.   rS   r0   s     €r1   r(   zCanaryDecoderMLP.__init__a   sf   ø€ Ý‰Œ×ÒÑÔÐØˆŒÝ# FÔ$5Ô6ˆÔÝ”9˜VÔ/°Ô1IÑJÔJˆŒÝ”9˜VÔ5°vÔ7IÑJÔJˆŒˆˆr2   Úhidden_statesr3   c                 ó„   — |                       |¦  «        }|                      |¦  «        }|                      |¦  «        }|S rE   )rq   rm   rr   )r.   rt   s     r1   rH   zCanaryDecoderMLP.forwardh   s=   € ØŸš Ñ/Ô/ˆØ×*Ò*¨=Ñ9Ô9ˆØŸš Ñ/Ô/ˆØÐr2   )rI   rJ   rK   r(   r8   rN   rH   rO   rP   s   @r1   rj   rj   `   sc   ø€ € € € € ðKð Kð Kð Kð Kð U¤\ð °e´lð ð ð ð ð ð ð ð r2   rj   rt   Ún_repr3   c                 ó¸   — | j         \  }}}}|dk    r| S | dd…dd…ddd…dd…f                              |||||¦  «        } |                      |||z  ||¦  «        S )zÔ
    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
    r   N)ÚshapeÚexpandÚreshape)rt   rv   ÚbatchÚnum_key_value_headsÚslenÚhead_dims         r1   Ú	repeat_kvr   o   s„   € ð
 2?Ô1DÑ.€EÐ  hØ�‚z€zØÐØ! ! ! ! Q Q Q¨¨a¨a¨a°°°Ð"2Ô3×:Ò:¸5ÐBUÐW\Ð^bÐdlÑmÔm€MØ× Ò  Ð(;¸eÑ(CÀTÈ8ÑTÔTÐTr2   ç        r]   ÚqueryÚkeyÚvalueÚattention_maskÚscalingÚdropoutÚkwargsc                 ó  — t          || j        ¦  «        }t          || j        ¦  «        }	t          j        ||                     dd¦  «        ¦  «        |z  }
|�|
|z   }
t
          j                             |
dt          j        ¬¦  «         	                    |j
        ¦  «        }
t
          j                             |
|| j        ¬¦  «        }
t          j        |
|	¦  «        }|                     dd¦  «                             ¦   «         }||
fS )Nr   r   éÿÿÿÿ)r5   Údtype)ÚpÚtrainingr   )r   Únum_key_value_groupsr8   ÚmatmulÚ	transposer   Ú
functionalÚsoftmaxÚfloat32ÚtorŠ   r†   rŒ   Ú
contiguous)r]   r�   r‚   rƒ   r„   r…   r†   r‡   Ú
key_statesÚvalue_statesÚattn_weightsÚattn_outputs               r1   Úeager_attention_forwardr™   {   sé   € õ ˜3 Ô ;Ñ<Ô<€JÝ˜U FÔ$?Ñ@Ô@€Lå”<  z×';Ò';¸A¸qÑ'AÔ'AÑBÔBÀWÑL€LØÐ!Ø# nÑ4ˆå”=×(Ò(¨¸2ÅUÄ]Ð(ÑSÔS×VÒVÐW\ÔWbÑcÔc€LÝ”=×(Ò(¨¸È6Ì?Ð(Ñ[Ô[€LÝ”,˜|¨\Ñ:Ô:€KØ×'Ò'¨¨1Ñ-Ô-×8Ò8Ñ:Ô:€Kà˜Ð$Ð$r2   c                   óZ   ‡ — e Zd Zdedefˆ fd„Z	 d	dej        dej        dedz  fd„Z	ˆ xZ
S )
ÚCanarySelfAttentionrS   Ú	layer_idxc                 ó®  •— t          ¦   «                              ¦   «          || _        || _        t	          |d|j        |j        z  ¦  «        | _        |j        |j        z  | _	        | j        dz  | _
        |j        | _        d| _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        | j        z  |j        |j        ¬¦  «        | _        d S )Nr~   ç      à¿T©Úbias©r'   r(   rS   rœ   Úgetattrro   Únum_attention_headsr~   r|   r�   r…   Úattention_dropoutÚ	is_causalr   rn   Úattention_biasÚq_projÚk_projÚv_projÚo_proj©r.   rS   rœ   r0   s      €r1   r(   zCanarySelfAttention.__init__–   sB  ø€ Ý‰Œ×ÒÑÔÐØˆŒØ"ˆŒÝ ¨
°FÔ4FÈ&ÔJdÑ4dÑeÔeˆŒØ$*Ô$>À&ÔB\Ñ$\ˆÔ!Ø”} dÑ*ˆŒØ!'Ô!9ˆÔØˆŒå”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ&¨¬Ñ6¸Ô8JÐQWÔQfð
ñ 
ô 
ˆŒˆˆr2   Nrt   r„   Úpast_key_valuesc                 ó  — |j         d d…         }g |¢d‘| j        ‘R }|                      |¦  «        }|                      |¦  «        }|                      |¦  «        }	|                     |¦  «                             dd¦  «        }|                     |¦  «                             dd¦  «        }|	                     |¦  «                             dd¦  «        }	|�&|j        }|                     ||	| j	        ¦  «        \  }}	t          j        | j        j        t          ¦  «        }
 |
| |||	|f| j        sdn| j        | j        dœ|¤Ž\  }} |j        g |¢d‘R Ž                      ¦   «         }|                      |¦  «        }||fS )Nr‰   r   r   r€   ©r†   r…   )rx   r~   r§   r¨   r©   Úviewr�   Úself_attention_cacheÚupdaterœ   r   Úget_interfacerS   Ú_attn_implementationr™   rŒ   r¤   r…   rz   r”   rª   )r.   rt   r„   r¬   r‡   Úinput_shapeÚhidden_shapeÚquery_statesr•   r–   Úattention_interfacer˜   r—   s                r1   rH   zCanarySelfAttention.forward­   s¯  € ð $Ô)¨#¨2¨#Ô.ˆØ8˜Ð8 bÐ8¨$¬-Ð8Ð8ˆà—{’{ =Ñ1Ô1ˆØ—[’[ Ñ/Ô/ˆ
Ø—{’{ =Ñ1Ô1ˆà#×(Ò(¨Ñ6Ô6×@Ò@ÀÀAÑFÔFˆØ—_’_ \Ñ2Ô2×<Ò<¸QÀÑBÔBˆ
Ø#×(Ò(¨Ñ6Ô6×@Ò@ÀÀAÑFÔFˆàÐ&Ø-ÔBˆOØ'6×'=Ò'=¸jÈ,ÐX\ÔXfÑ'gÔ'gÑ$ˆJ˜å(?Ô(MØŒKÔ,Õ.Eñ)
ô )
Ðð %8Ð$7ØØØØØð	%
ð  $œ}ÐH�C�C°$Ô2HØ”Lð	%
ð 	%
ð ð	%
ð 	%
Ñ!ˆ�\ð *�kÔ)Ð;¨;Ð;¸Ð;Ð;Ð;×FÒFÑHÔHˆØ—k’k +Ñ.Ô.ˆØ˜LÐ(Ð(r2   rE   )rI   rJ   rK   r   rM   r(   r8   rN   r   rH   rO   rP   s   @r1   r›   r›   •   s�   ø€ € € € € ð
˜|ð 
¸ð 
ð 
ð 
ð 
ð 
ð 
ð6 )-ð	')ð ')à”|ð')ð œð')ð  ™ð	')ð ')ð ')ð ')ð ')ð ')ð ')ð ')r2   r›   c                   óˆ   ‡ — e Zd Zdedefˆ fd„Z	 	 	 ddej        dej        dz  dej        dz  dedz  d	e	e
         f
d
„Zˆ xZS )ÚCanaryCrossAttentionrS   rœ   c                 ó®  •— t          ¦   «                              ¦   «          || _        || _        t	          |d|j        |j        z  ¦  «        | _        |j        |j        z  | _	        | j        dz  | _
        |j        | _        d| _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        | j        z  |j        |j        ¬¦  «        | _        d S )Nr~   rž   FrŸ   r¡   r«   s      €r1   r(   zCanaryCrossAttention.__init__Ù   sB  ø€ Ý‰Œ×ÒÑÔÐØˆŒØ"ˆŒÝ ¨
°FÔ4FÈ&ÔJdÑ4dÑeÔeˆŒØ$*Ô$>À&ÔB\Ñ$\ˆÔ!Ø”} dÑ*ˆŒØ!'Ô!9ˆÔØˆŒå”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ&¨¬Ñ6¸Ô8JÐQWÔQfð
ñ 
ô 
ˆŒˆˆr2   Nrt   Úencoder_hidden_statesr„   r¬   r‡   c                 óÔ  — |j         d d…         \  }}|j         d         }||d| j        f}	||d| j        f}
 |                      |¦  «        j        |	Ž                      dd¦  «        }|�|j                             | j        ¦  «        nd}|�;|r9|j        j	        | j                 j
        }|j        j	        | j                 j        }n— |                      |¦  «        j        |
Ž                      dd¦  «        } |                      |¦  «        j        |
Ž                      dd¦  «        }|�3|j                             ||| j        ¦  «        \  }}d|j        | j        <   t          j        | j        j        t&          ¦  «        } || ||||f| j        sdn| j        | j        dœ|¤Ž\  }}|                     ||d¦  «                             ¦   «         }|                      |¦  «        }||fS )Nr‰   r   r   FTr€   r®   )rx   r~   r§   r¯   r�   Ú
is_updatedÚgetrœ   Úcross_attention_cacheÚlayersÚkeysÚvaluesr¨   r©   r±   r   r²   rS   r³   r™   rŒ   r¤   r…   rz   r”   rª   )r.   rt   r»   r„   r¬   r‡   ÚbszÚtgt_lenÚsrc_lenÚq_input_shapeÚkv_input_shaper¶   r½   r•   r–   r·   r˜   r—   s                     r1   rH   zCanaryCrossAttention.forwardð   s  € ð %Ô*¨3¨B¨3Ô/‰ˆˆWØ'Ô-¨aÔ0ˆà˜g r¨4¬=Ð9ˆØ˜w¨¨D¬MÐ:ˆð 7�t—{’{ =Ñ1Ô1Ô6¸ÐF×PÒPÐQRÐTUÑVÔVˆàGVÐGb�_Ô/×3Ò3°D´NÑCÔCÐCÐhmˆ
ØÐ&¨:Ð&à(Ô>ÔEÀdÄnÔUÔZˆJØ*Ô@ÔGÈÌÔWÔ^ˆLˆLà@˜ŸšÐ%:Ñ;Ô;Ô@À.ÐQ×[Ò[Ð\]Ð_`ÑaÔaˆJØB˜4Ÿ;š;Ð'<Ñ=Ô=ÔBÀNÐS×]Ò]Ð^_ÐabÑcÔcˆLàÐ*à+:Ô+P×+WÒ+WØ ¨d¬nñ,ô ,Ñ(�
˜Lð >B�Ô*¨4¬>Ñ:å(?Ô(MØŒKÔ,Õ.Eñ)
ô )
Ðð %8Ð$7ØØØØØð	%
ð  $œ}ÐH�C�C°$Ô2HØ”Lð	%
ð 	%
ð ð	%
ð 	%
Ñ!ˆ�\ð "×)Ò)¨#¨w¸Ñ;Ô;×FÒFÑHÔHˆØ—k’k +Ñ.Ô.ˆØ˜LÐ(Ð(r2   )NNN)rI   rJ   rK   r   rM   r(   r8   rN   r   r   r   rH   rO   rP   s   @r1   r¹   r¹   Ø   s¹   ø€ € € € € ð
˜|ð 
¸ð 
ð 
ð 
ð 
ð 
ð 
ð4 6:Ø.2Ø(,ð3)ð 3)à”|ð3)ð  %œ|¨dÑ2ð3)ð œ tÑ+ð	3)ð
  ™ð3)ð Ð+Ô,ð3)ð 3)ð 3)ð 3)ð 3)ð 3)ð 3)ð 3)r2   r¹   c                   ó
  ‡ — e Zd Zdˆ fd„	Z	 	 	 	 	 	 ddej        dej        dz  dej        dz  dej        dz  dej        dz  dej        dz  d	edz  d
ee	         de
ej        e
ej        ej        f         dz  f         fd„Zˆ xZS )rW   Nc                 ó|  •— t          ¦   «                              ¦   «          t          ||¬¦  «        | _        t	          ||¬¦  «        | _        t          |¦  «        | _        t          j	        |j
        ¦  «        | _        t          j	        |j
        ¦  «        | _        t          j	        |j
        ¦  «        | _        d S )N)rS   rœ   )r'   r(   r›   Ú	self_attnr¹   Úencoder_attnrj   Úmlpr   Ú	LayerNormro   Úinput_layernormÚpost_attention_layernormÚfinal_layernormr«   s      €r1   r(   zCanaryDecoderLayer.__init__'  s–   ø€ Ý‰Œ×ÒÑÔÐÝ,°FÀiÐPÑPÔPˆŒÝ0¸È)ÐTÑTÔTˆÔå# FÑ+Ô+ˆŒÝ!œ|¨FÔ,>Ñ?Ô?ˆÔÝ(*¬°VÔ5GÑ(HÔ(HˆÔ%Ý!œ|¨FÔ,>Ñ?Ô?ˆÔÐÐr2   rt   r„   r»   Úencoder_attention_maskrC   Úencoder_position_idsr¬   r‡   r3   c                 ó>  — |}	|                       |¦  «        } | j        d||||dœ|¤Ž\  }}
|	|z   }|�8|}	|                      |¦  «        }|                      ||||¬¦  «        \  }}
|	|z   }|}	|                      |¦  «        }|                      |¦  «        }|	|z   }|S )N)rt   r„   rC   r¬   )rt   r»   r„   r¬   © )rÎ   rÊ   rÏ   rË   rÐ   rÌ   )r.   rt   r„   r»   rÑ   rC   rÒ   r¬   r‡   ÚresidualÚ_s              r1   rH   zCanaryDecoderLayer.forward1  sî   € ð !ˆØ×,Ò,¨]Ñ;Ô;ˆà)˜4œ>ð 
Ø'Ø)Ø%Ø+ð	
ð 
ð
 ð
ð 
Ñˆ�qð ! =Ñ0ˆà Ð,Ø$ˆHØ ×9Ò9¸-ÑHÔHˆMØ#×0Ò0Ø+Ø&;Ø5Ø /ð	  1ñ  ô  ÑˆM˜1ð % }Ñ4ˆMà ˆØ×,Ò,¨]Ñ;Ô;ˆØŸš Ñ/Ô/ˆØ  =Ñ0ˆØÐr2   rE   )NNNNNN)rI   rJ   rK   r(   r8   rN   Ú
LongTensorr   r   r   ÚtupleÚFloatTensorrH   rO   rP   s   @r1   rW   rW   &  s  ø€ € € € € ð@ð @ð @ð @ð @ð @ð /3Ø59Ø6:Ø04Ø8<Ø(,ð&ð &à”|ð&ð œ tÑ+ð&ð  %œ|¨dÑ2ð	&ð
 !&¤¨tÑ 3ð&ð Ô&¨Ñ-ð&ð $Ô.°Ñ5ð&ð  ™ð&ð Ð+Ô,ð&ð 
ˆuÔ  %¨Ô(9¸5Ô;LÐ(LÔ"MÐPTÑ"TÐTÔ	Uð&ð &ð &ð &ð &ð &ð &ð &r2   rW   c                   ó`  ‡ — e Zd ZU dZ eedd¬¦  «        e eedd¬¦  «        dœZe	e
d<   de	fˆ fd„Zeee	 	 	 	 	 	 	 	 ddej        d	z  d
ej        d	z  dej        d	z  ded	z  dej        d	z  ded	z  dej        d	z  dej        d	z  dee         deez  fd„¦   «         ¦   «         ¦   «         Zˆ xZS )ÚCanaryDecoderÚ	input_idsr   rÊ   )ÚindexÚ
layer_namerË   )Ú
attentionsrt   Úcross_attentionsrS   c                 ó*  •‡— t          ¦   «                              ‰¦  «         ‰j        | _        ‰j        | _        t          j        ‰j        ‰j        | j        ¦  «        | _        t          j	        ˆfd„t          ‰j        ¦  «        D ¦   «         ¦  «        | _        t          j        ‰j        ¦  «        | _        d| _        t!          ‰j        ‰j        ¦  «        | _        t          j        ‰j        ¦  «        | _        |                      ¦   «          d S )Nc                 ó0   •— g | ]}t          ‰|¦  «        ‘ŒS rÔ   )rW   )Ú.0ÚidxrS   s     €r1   ú
<listcomp>z*CanaryDecoder.__init__.<locals>.<listcomp>j  s$   ø€ Ð$pÐ$pÐ$pÈÕ%7¸ÀÑ%DÔ%DÐ$pÐ$pÐ$pr2   F)r'   r(   Úpad_token_idÚpadding_idxÚ
vocab_sizer   Ú	Embeddingro   Úembed_tokensÚ
ModuleListÚrangeÚnum_hidden_layersrÀ   rÍ   ÚnormÚgradient_checkpointingr"   Úmax_position_embeddingsÚpos_embÚembedding_layernormÚ	post_initrs   s    `€r1   r(   zCanaryDecoder.__init__d  så   øø€ Ý‰Œ×Ò˜Ñ Ô Ð Ø!Ô.ˆÔØ Ô+ˆŒåœL¨Ô):¸FÔ<NÐPTÔP`ÑaÔaˆÔÝ”mÐ$pÐ$pÐ$pÐ$pÕPUÐV\ÔVnÑPoÔPoÐ$pÑ$pÔ$pÑqÔqˆŒÝ”L Ô!3Ñ4Ô4ˆŒ	Ø&+ˆÔ#Ý0°Ô1OÐQWÔQcÑdÔdˆŒÝ#%¤<°Ô0BÑ#CÔ#CˆÔ ð 	�ŠÑÔÐÐÐr2   Nr„   rC   r¬   Úinputs_embedsÚ	use_cacher»   rÑ   r‡   r3   c	                 ó2  — |du |duz  rt          d¦  «        ‚|€|                      |¦  «        }|r8|€6t          t          | j        ¬¦  «        t          | j        ¬¦  «        ¦  «        }|€V|�|                     ¦   «         nd}
t          j        |j        d         |j	        ¬¦  «        |
z   }| 
                    d¦  «        }|                      |                     d¦  «        ¦  «        }|                     |j	        |j        ¬¦  «        }|                      ||z   ¦  «        }t!          | j        ||||¬¦  «        }t#          | j        |||¬	¦  «        }|}| j        D ]} ||||f|||d
œ|	¤Ž}Œ|                      |¦  «        }t)          ||r|nd¬¦  «        S )a¤  
        encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
            Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
            of the decoder.
        encoder_attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
            Mask to avoid performing attention on padding indices in `encoder_hidden_states`. 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)
        Nz:You must specify exactly one of input_ids or inputs_embeds)rS   r   r   )Údevice)r÷   rŠ   )rS   rô   r„   r¬   rC   )rS   rô   r„   r»   )rÑ   rC   r¬   )Úlast_hidden_stater¬   )r*   rê   r
   r	   rS   Úget_seq_lengthr8   r:   rx   r÷   Ú	unsqueezerñ   Úsqueezer“   rŠ   rò   r   r   rÀ   rî   r   )r.   rÜ   r„   rC   r¬   rô   rõ   r»   rÑ   r‡   Úpast_seen_tokensrñ   Úcausal_maskrt   Údecoder_layers                  r1   rH   zCanaryDecoder.forwards  s  € ð2 ˜Ð -°tÐ";Ñ<ð 	[ÝÐYÑZÔZÐZàÐ Ø ×-Ò-¨iÑ8Ô8ˆMàð 	v˜Ð0Ý1µ,ÀdÄkÐ2RÑ2RÔ2RÕT`ÐhlÔhsÐTtÑTtÔTtÑuÔuˆOàÐØCRÐC^˜×=Ò=Ñ?Ô?Ð?ÐdeÐÝ œ<¨Ô(;¸AÔ(>À}ÔG[Ð\Ñ\Ô\Ð_oÑoˆLØ'×1Ò1°!Ñ4Ô4ˆLð —,’,˜|×3Ò3°AÑ6Ô6Ñ7Ô7ˆØ—*’* MÔ$8ÀÔ@S�*ÑTÔTˆØ×0Ò0°ÀÑ1HÑIÔIˆå(Ø”;Ø'Ø)Ø+Ø%ð
ñ 
ô 
ˆõ ";Ø”;Ø'Ø1Ø"7ð	"
ñ "
ô "
Ðð &ˆØ!œ[ð 		ð 		ˆMØ)˜MØØØ%ðð (>Ø)Ø /ðð ð ðð ˆMˆMð Ÿ	š	 -Ñ0Ô0ˆå8Ø+Ø/8ÐB˜O˜O¸dð
ñ 
ô 
ð 	
r2   )NNNNNNNN)rI   rJ   rK   r`   r   r›   rW   r¹   Ú_can_record_outputsr    r^   r(   r   r   r   r8   r×   rN   r   rÙ   Úboolr   r   rØ   r   rH   rO   rP   s   @r1   rÛ   rÛ   Z  s   ø€ € € € € € à!€Oà$�nÐ%8ÀÈkÐZÑZÔZØ+Ø*˜NÐ+?ÀqÐUcÐdÑdÔdðð Ðð
  ÐÐÑðÐ2ð ð ð ð ð ð ð  ØØð .2Ø.2Ø04Ø(,Ø26Ø!%Ø:>Ø6:ðH
ð H
àÔ# dÑ*ðH
ð œ tÑ+ðH
ð Ô&¨Ñ-ð	H
ð
  ™ðH
ð Ô(¨4Ñ/ðH
ð ˜$‘;ðH
ð  %Ô0°4Ñ7ðH
ð !&¤¨tÑ 3ðH
ð Ð+Ô,ðH
ð 
Ð:Ñ	:ðH
ð H
ð H
ñ „^ñ „_ñ  ÔðH
ð H
ð H
ð H
ð H
r2   rÛ   z�
    The bare Canary model (FastConformer encoder + Transformer decoder) outputting raw hidden-states without any
    specific head on top.
    )Úcustom_introc                   ó\  ‡ — e Zd Zdefˆ fd„Zd„ Zd„ Zd„ Zd„ Ze	e
	 	 	 	 	 	 	 	 	 ddej        dz  d	ej        dz  d
ej        dz  dej        dz  deeej                          dz  dedz  deej                 dz  deej                 dz  dedz  dee         defd„¦   «         ¦   «         Zˆ xZS )ÚCanaryModelrS   c                 óà   •— t          ¦   «                              |¦  «         t          j        |j        ¦  «        | _        t          |j        ¦  «        | _        |  	                    ¦   «          d S rE   )
r'   r(   r   Úfrom_configÚencoder_configÚencoderrÛ   Údecoder_configÚdecoderró   rs   s     €r1   r(   zCanaryModel.__init__È  sX   ø€ Ý‰Œ×Ò˜Ñ Ô Ð Ý Ô,¨VÔ-BÑCÔCˆŒÝ$ VÔ%:Ñ;Ô;ˆŒà�ŠÑÔÐÐÐr2   c                 ó   — | j         j        S rE   ©r	  rê   ©r.   s    r1   Úget_input_embeddingsz CanaryModel.get_input_embeddingsÏ  s   € ØŒ|Ô(Ð(r2   c                 ó   — || j         _        d S rE   r  )r.   rƒ   s     r1   Úset_input_embeddingsz CanaryModel.set_input_embeddingsÒ  s   € Ø$)ˆŒÔ!Ð!Ð!r2   c                 ó8   — | j                              ¦   «          dS )z¨
        Calling this function will disable the gradient computation for the Canary encoder so that its parameters will
        not be updated during training.
        N)r  Ú_freeze_parametersr  s    r1   Úfreeze_encoderzCanaryModel.freeze_encoderÕ  s   € ð
 	Œ×'Ò'Ñ)Ô)Ð)Ð)Ð)r2   c                 ó    — t          d¦  «        ‚)z¢
        Masks extracted features along time axis and/or along feature axis according to
        [SpecAugment](https://huggingface.co/papers/1904.08779).
        zNot needed for Canary)ÚAttributeErrorr  s    r1   Ú_mask_input_featuresz CanaryModel._mask_input_featuresÜ  s   € õ
 Ð4Ñ5Ô5Ð5r2   NrU   r„   Údecoder_input_idsÚdecoder_attention_maskÚencoder_outputsr¬   Údecoder_inputs_embedsÚdecoder_position_idsrõ   r‡   r3   c
                 óä   — |€ | j         |fd|i|
¤Ž} | j        d|||j        |j        ||||	dœ|
¤Ž}t	          |j        |j        |j        |j        |j        |j        |j        |j        ¬¦  «        S )a  
        input_features (`torch.FloatTensor` of shape `(batch_size, audio_length)`):
            Float values of the raw speech waveform. Raw speech waveform 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_features`, the [`AutoFeatureExtractor`] should be used for padding
            and conversion into a tensor of type `torch.FloatTensor`.
        decoder_position_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`):
            Indices of positions of each input sequence tokens in the position embeddings.
            Used to calculate the position embeddings up to `config.decoder_config.max_position_embeddings`

        Example:

        ```python
        >>> import torch
        >>> from transformers import AutoFeatureExtractor, CanaryModel
        >>> from datasets import load_dataset

        >>> model = CanaryModel.from_pretrained("UsefulSensors/canary-tiny")
        >>> feature_extractor = AutoFeatureExtractor.from_pretrained("UsefulSensors/canary-tiny")
        >>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
        >>> inputs = feature_extractor(ds[0]["audio"]["array"], return_tensors="pt")
        >>> input_features = inputs.input_features
        >>> decoder_input_ids = torch.tensor([[1, 1]]) * model.config.decoder_start_token_id
        >>> last_hidden_state = model(input_features, decoder_input_ids=decoder_input_ids).last_hidden_state
        >>> list(last_hidden_state.shape)
        [1, 2, 288]
        ```
        Nr„   )rÜ   r„   r»   rÑ   r¬   rô   rC   rõ   )rø   r¬   Údecoder_hidden_statesÚdecoder_attentionsrà   Úencoder_last_hidden_stater»   Úencoder_attentionsrÔ   )	r  r	  rø   r„   r   r¬   rt   rß   rà   )r.   rU   r„   r  r  r  r¬   r  r  rõ   r‡   Údecoder_outputss               r1   rH   zCanaryModel.forwardã  s¾   € ð\ Ð"Ø/;¨t¬|¸NÐ/tÐ/tÐ[iÐ/tÐmsÐ/tÐ/tˆOàEQÀTÄ\ð 
F
Ø'Ø1Ø"1Ô"CØ#2Ô#AØ+Ø/Ø-Øð
F
ð 
F
ð ð
F
ð 
F
ˆõ "Ø-Ô?Ø+Ô;Ø"1Ô"?Ø.Ô9Ø,Ô=Ø&5Ô&GØ"1Ô"?Ø.Ô9ð	
ñ 	
ô 	
ð 		
r2   )	NNNNNNNNN)rI   rJ   rK   r   r(   r  r  r  r  r   r   r8   rÙ   r×   rØ   r
   r   r   r   r   rH   rO   rP   s   @r1   r  r  Á  s¨  ø€ € € € € ð˜|ð ð ð ð ð ð ð)ð )ð )ð*ð *ð *ð*ð *ð *ð6ð 6ð 6ð Øð 48Ø26Ø59Ø:>ØBFØ6:ØAEØ?CØ!%ðD
ð D
àÔ)¨DÑ0ðD
ð Ô(¨4Ñ/ðD
ð !Ô+¨dÑ2ð	D
ð
 !&Ô 0°4Ñ 7ðD
ð ˜u UÔ%6Ô7Ô8¸4Ñ?ðD
ð -¨tÑ3ðD
ð  % UÔ%6Ô7¸$Ñ>ðD
ð $ EÔ$4Ô5¸Ñ<ðD
ð ˜$‘;ðD
ð Ð+Ô,ðD
ð 
ðD
ð D
ð D
ñ „^ñ ÔðD
ð D
ð D
ð D
ð D
r2   r  rÜ   ræ   Údecoder_start_token_idc                 óô   — |                       | j        ¦  «        }| dd…dd…f                              ¦   «         |dd…dd…f<   ||dd…df<   |€t          d¦  «        ‚|                     |dk    |¦  «         |S )z1
    Shift input ids one token to the right.
    Nr‰   r   r   z1self.model.config.pad_token_id has to be defined.iœÿÿÿ)Ú	new_zerosrx   Úcloner*   Úmasked_fill_)rÜ   ræ   r!  Úshifted_input_idss       r1   Úshift_tokens_rightr'  ,  s˜   € ð "×+Ò+¨I¬OÑ<Ô<ÐØ(¨¨¨¨C¨R¨C¨Ô0×6Ò6Ñ8Ô8Ð�a�a�a˜˜˜�eÑØ4Ð�a�a�a˜�dÑàÐÝÐLÑMÔMÐMà×"Ò"Ð#4¸Ò#<¸lÑKÔKÐKàÐr2   z—
    The Canary model with a language modeling head. Can be used for multilingual automatic speech recognition and
    speech-to-text translation.
    c                   ó”  ‡ — e Zd ZddiZdefˆ fd„Zd„ Zd„ Zdej	        fd„Z
ee	 	 	 	 	 	 	 	 	 	 dd
ej        d	z  dej        d	z  dej        d	z  dej        d	z  deeej                          d	z  ded	z  deej                 d	z  deej                 d	z  ded	z  dej        d	z  dee         defd„¦   «         ¦   «         Zd	dœˆ fd„
Zˆ xZS )ÚCanaryForConditionalGenerationzproj_out.weightz!model.decoder.embed_tokens.weightrS   c                 óú   •— t          ¦   «                              |¦  «         t          |¦  «        | _        t	          j        |j        j        |j        j        d¬¦  «        | _	        |  
                    ¦   «          d S )NTrŸ   )r'   r(   r  rT   r   rn   r  ro   rè   Úproj_outró   rs   s     €r1   r(   z'CanaryForConditionalGeneration.__init__E  sg   ø€ Ý‰Œ×Ò˜Ñ Ô Ð Ý  Ñ(Ô(ˆŒ
Ýœ	 &Ô"7Ô"CÀVÔEZÔEeÐlpÐqÑqÔqˆŒð 	�ŠÑÔÐÐÐr2   c                 ó   — | j         S rE   ©r+  r  s    r1   Úget_output_embeddingsz4CanaryForConditionalGeneration.get_output_embeddingsM  s
   € ØŒ}Ðr2   c                 ó   — || _         d S rE   r-  )r.   Únew_embeddingss     r1   Úset_output_embeddingsz4CanaryForConditionalGeneration.set_output_embeddingsP  s   € Ø&ˆŒˆˆr2   r3   c                 ó4   — | j                              ¦   «         S rE   )rT   r  r  s    r1   r  z3CanaryForConditionalGeneration.get_input_embeddingsS  s   € ØŒz×.Ò.Ñ0Ô0Ð0r2   NrU   r„   r  r  r  r¬   r  r  rõ   Úlabelsr‡   c                 óœ  — |
�)|€'|€%t          |
| j        j        | j        j        ¦  «        } | j        |f||||||||	dœ|¤Ž}|                      |j        ¦  «        }d}|
�2|                     d|
¦  «        } | j        d||
| j        j	        |dœ|¤Ž}t          |||j        |j        |j        |j        |j        |j        |j        ¬¦	  «	        S )a.  
        input_features (`torch.FloatTensor` of shape `(batch_size, audio_length)`):
            Float values of the raw speech waveform. Raw speech waveform 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_features`, the [`AutoFeatureExtractor`] should be used for padding
            and conversion into a tensor of type `torch.FloatTensor`.
        decoder_position_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`):
            Indices of positions of each input sequence tokens in the position embeddings.
            Used to calculate the position embeddings up to `config.decoder_config.max_position_embeddings`

        Example:

        ```python
        >>> import torch
        >>> from transformers import AutoProcessor, CanaryForConditionalGeneration
        >>> from datasets import load_dataset

        >>> processor = AutoProcessor.from_pretrained("UsefulSensors/canary-tiny")
        >>> model = CanaryForConditionalGeneration.from_pretrained("UsefulSensors/canary-tiny")

        >>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")

        >>> inputs = processor(ds[0]["audio"]["array"], return_tensors="pt")
        >>> input_features = inputs.input_features

        >>> generated_ids = model.generate(input_features, max_new_tokens=100)

        >>> transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
        >>> transcription
        'Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.'
        ```N)r„   r  r  r  r¬   r  r  rõ   Úshift_labels)Úlogitsr3  rè   r5  )	Úlossr6  r¬   r  r  rà   r  r»   r  rÔ   )r'  rS   ræ   r!  rT   r+  rø   ÚpopÚloss_functionrè   r   r¬   r  r  rà   r  r»   r  )r.   rU   r„   r  r  r  r¬   r  r  rõ   r3  r‡   Úoutputsr6  r7  r5  s                   r1   rH   z&CanaryForConditionalGeneration.forwardV  s2  € ðd ÐØ Ð(Ð-BÐ-JÝ$6Ø˜DœKÔ4°d´kÔ6Xñ%ô %Ð!ð '1 d¤jØð'
à)Ø/Ø+Ø#9Ø+Ø"7Ø!5Øð'
ð '
ð ð'
ð '
ˆð —’˜wÔ8Ñ9Ô9ˆàˆØÐØ!Ÿ:š: n°fÑ=Ô=ˆLØ%�4Ô%ð ØØØœ;Ô1Ø)ð	ð ð
 ðð ˆDõ ØØØ#Ô3Ø")Ô"?Ø&Ô9Ø$Ô5Ø&-Ô&GØ")Ô"?Ø&Ô9ð

ñ 

ô 

ð 
	
r2   )Úaudio_chunk_indexc                ó6   •—  t          ¦   «         j        |i |¤ŽS rE   )r'   Úprepare_inputs_for_generation)r.   r;  Úargsr‡   r0   s       €r1   r=  z<CanaryForConditionalGeneration.prepare_inputs_for_generation³  s   ø€ à4�u‰wŒwÔ4°dÐE¸fÐEÐEÐEr2   )
NNNNNNNNNN)rI   rJ   rK   Ú_tied_weights_keysr   r(   r.  r1  r   ÚModuler  r   r   r8   rÙ   r×   rØ   r
   r   r   r   r   rH   r=  rO   rP   s   @r1   r)  r)  <  sý  ø€ € € € € ð ,Ð-PÐQÐð˜|ð ð ð ð ð ð ðð ð ð'ð 'ð 'ð1 b¤ið 1ð 1ð 1ð 1ð Øð 48Ø26Ø59Ø:>ØBFØ6:ØAEØ?CØ!%Ø*.ðY
ð Y
àÔ)¨DÑ0ðY
ð Ô(¨4Ñ/ðY
ð !Ô+¨dÑ2ð	Y
ð
 !&Ô 0°4Ñ 7ðY
ð ˜u UÔ%6Ô7Ô8¸4Ñ?ðY
ð -¨tÑ3ðY
ð  % UÔ%6Ô7¸$Ñ>ðY
ð $ EÔ$4Ô5¸Ñ<ðY
ð ˜$‘;ðY
ð Ô  4Ñ'ðY
ð Ð+Ô,ðY
ð 
ðY
ð Y
ð Y
ñ „^ñ ÔðY
ðv FJð Fð Fð Fð Fð Fð Fð Fð Fð Fð Fð Fr2   r)  )r)  r  rR   )r€   )@rF   Úcollections.abcr   Únumpyr6   r8   r   Ú r   r[   Úactivationsr   Úcache_utilsr   r	   r
   Ú
generationr   Úmasking_utilsr   r   Úmodeling_layersr   Úmodeling_outputsr   r   r   r   Úmodeling_utilsr   r   Úprocessing_utilsr   Úutilsr   r   Úutils.genericr   r   Úutils.output_capturingr   r   Úauto.modeling_autor   Úconfiguration_canaryr   r    r@  r"   rR   rj   rN   rM   r   r;   r™   r›   r¹   rW   rÛ   r  r'  r)  Ú__all__rÔ   r2   r1   ú<module>rR     s»  ðð* €€€Ø $Ð $Ð $Ð $Ð $Ð $à Ð Ð Ð Ø €€€Ø Ð Ð Ð Ð Ð à &Ð &Ð &Ð &Ð &Ð &Ø !Ð !Ð !Ð !Ð !Ð !Ø CÐ CÐ CÐ CÐ CÐ CÐ CÐ CÐ CÐ CØ )Ð )Ð )Ð )Ð )Ð )Ø JÐ JÐ JÐ JÐ JÐ JÐ JÐ JØ 9Ð 9Ð 9Ð 9Ð 9Ð 9ðð ð ð ð ð ð ð ð ð ð ð ð GÐ FÐ FÐ FÐ FÐ FÐ FÐ FØ &Ð &Ð &Ð &Ð &Ð &Ø 7Ð 7Ð 7Ð 7Ð 7Ð 7Ð 7Ð 7Ø IÐ IÐ IÐ IÐ IÐ IÐ IÐ IØ EÐ EÐ EÐ EÐ EÐ EÐ EÐ EØ *Ð *Ð *Ð *Ð *Ð *Ø CÐ CÐ CÐ CÐ CÐ CÐ CÐ CðRð Rð Rð Rð R ¤	ñ Rô Rð Rð4 ðlð lð lð lð l˜Oñ lô lñ „ðlð(ð ð ð ð �r”yñ ô ð ð	U˜Uœ\ð 	U°#ð 	U¸%¼,ð 	Uð 	Uð 	Uð 	Uð& ð%ð %ØŒIð%àŒ<ð%ð 
Œð%ð Œ<ð	%ð
 ”L 4Ñ'ð%ð ð%ð ð%ð Ð'Ô(ð%ð %ð %ð %ð4?)ð ?)ð ?)ð ?)ð ?)˜"œ)ñ ?)ô ?)ð ?)ðFK)ð K)ð K)ð K)ð K)˜2œ9ñ K)ô K)ð K)ð\1ð 1ð 1ð 1ð 1Ð3ñ 1ô 1ð 1ðh ðc
ð c
ð c
ð c
ð c
Ð)ñ c
ô c
ñ „ðc
ðL €ððñ ô ðb
ð b
ð b
ð b
ð b
Ð'ñ b
ô b
ñô ðb
ðJ %¤,ð ¸cð Ð[^ð ð ð ð ð  €ððñ ô ðsFð sFð sFð sFð sFÐ%:¸Oñ sFô sFñô ðsFðl UÐ
TÐ
T€€€r2   