§
    ‚Štj]Ž  ã                   ó  — d dl Z d dlmZ d dlmZmZ d dlZd dlmc m	Z
 d dlmZ ddl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 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$Z$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l0m1Z1m2Z2 ddl3m4Z4m5Z5 ddl6m7Z7  e/¦   «         rd dl8m9Z9  ed¦  «         G d„ dej:        ¦  «        ¦   «         Z; G d„ dej:        ¦  «        Z<d„ Z= ed¦  «        dId„¦   «         Z>d ej?        d!e@d"ej?        fd#„ZA	 dJd%ej:        d&ej?        d'ej?        d(ej?        d)ej?        dz  d*eBd+eBd,e*e,         fd-„ZC	 	 dKd%ej:        d&ej?        d'ej?        d(ej?        d)eej?        d.f         d*eBdz  d/eBdz  d"eDej?        ej?        f         fd0„ZE e'¦   «         ZFeEeFd1<    G d2„ d3ej:        ¦  «        ZG G d4„ d5ej:        ¦  «        ZH G d6„ d7ej:        ¦  «        ZI G d8„ d9e¦  «        ZJe- G d:„ d;e(¦  «        ¦   «         ZKe- G d<„ d=eK¦  «        ¦   «         ZL	 	 	 	 dLd?ej?        eDej?                 z  dz  d@e@dz  dAe@dz  dBe@d)ej?        dz  d"ej?        e@z  fdC„ZMe- G dD„ dEeKe¦  «        ¦   «         ZN G dF„ dGeeK¦  «        ZOg dH¢ZPdS )Mé    N)ÚCallable)ÚOptionalÚUnion)Únné   )Úinitialization)ÚACT2FN)ÚCacheÚDynamicCache)ÚGenerationMixin)Úuse_kernel_forward_from_hubÚuse_kernel_func_from_hub)Úcompile_friendly_flex_attention)Úcreate_causal_maskÚ!create_sliding_window_causal_mask)Ú GenericForSequenceClassificationÚGradientCheckpointingLayer)ÚMoeCausalLMOutputWithPastÚMoeModelOutputWithPast)ÚROPE_INIT_FUNCTIONSÚdynamic_rope_update)ÚAttentionInterfaceÚPreTrainedModel)ÚUnpack)ÚTransformersKwargsÚauto_docstringÚcan_return_tupleÚis_torch_flex_attn_available)Úmaybe_autocastÚmerge_with_config_defaults)ÚOutputRecorderÚcapture_outputsé   )Ú
DogeConfig)Ú	BlockMaskÚRMSNormc                   óT   ‡ — e Zd Zd	deddfˆ fd„Zdej        dej        fd„Zd„ Zˆ xZ	S )
ÚDogeRMSNormç�íµ ÷Æ°>ÚepsÚreturnNc                 ó¬   •— t          ¦   «                              ¦   «          t          j        t	          j        |¦  «        ¦  «        | _        || _        dS )z:
        DogeRMSNorm is equivalent to T5LayerNorm
        N)ÚsuperÚ__init__r   Ú	ParameterÚtorchÚonesÚweightÚvariance_epsilon)ÚselfÚhidden_sizer*   Ú	__class__s      €úd/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/transformers/models/doge/modeling_doge.pyr.   zDogeRMSNorm.__init__7   sD   ø€ õ 	‰Œ×ÒÑÔÐÝ”l¥5¤:¨kÑ#:Ô#:Ñ;Ô;ˆŒØ #ˆÔÐÐó    Úhidden_statesc                 ó  — |j         }|                     t          j        ¦  «        }|                     d¦  «                             dd¬¦  «        }|t          j        || j        z   ¦  «        z  }| j        |                     |¦  «        z  S )Né   éÿÿÿÿT)Úkeepdim)	ÚdtypeÚtor0   Úfloat32ÚpowÚmeanÚrsqrtr3   r2   )r4   r9   Úinput_dtypeÚvariances       r7   ÚforwardzDogeRMSNorm.forward?   s|   € Ø#Ô)ˆØ%×(Ò(­¬Ñ7Ô7ˆØ ×$Ò$ QÑ'Ô'×,Ò,¨R¸Ð,Ñ>Ô>ˆØ%­¬°H¸tÔ?TÑ4TÑ(UÔ(UÑUˆØŒ{˜]×-Ò-¨kÑ:Ô:Ñ:Ð:r8   c                 óH   — t          | j        j        ¦  «        › d| j        › �S )Nz, eps=)Útupler2   Úshaper3   )r4   s    r7   Ú
extra_reprzDogeRMSNorm.extra_reprF   s&   € Ý˜œÔ)Ñ*Ô*ÐIÐI°$Ô2GÐIÐIÐIr8   )r)   )
Ú__name__Ú
__module__Ú__qualname__Úfloatr.   r0   ÚTensorrF   rJ   Ú__classcell__©r6   s   @r7   r(   r(   5   sŒ   ø€ € € € € ð$ð $¨ð $¸$ð $ð $ð $ð $ð $ð $ð; U¤\ð ;°e´lð ;ð ;ð ;ð ;ðJð Jð Jð Jð Jð Jð Jr8   r(   c                   óÔ   ‡ — e Zd ZU ej        ed<   ddefˆ fd„Ze	 	 	 ddedz  de	d         de
dz  ded	ef         fd
„¦   «         Z ej        ¦   «         ed„ ¦   «         ¦   «         Zˆ xZS )ÚDogeRotaryEmbeddingÚinv_freqNÚconfigc                 ó²  •— t          ¦   «                              ¦   «          |j        | _        |j        | _        || _        | j        j        d         | _        | j        }| j        dk    rt          | j                 } || j        |¦  «        \  }| _
        |                      d|d¬¦  «         |                      d|                     ¦   «         d¬¦  «         d S )NÚ	rope_typeÚdefaultrT   F)Ú
persistentÚoriginal_inv_freq)r-   r.   Úmax_position_embeddingsÚmax_seq_len_cachedÚoriginal_max_seq_lenrU   Úrope_parametersrW   Úcompute_default_rope_parametersr   Úattention_scalingÚregister_bufferÚclone)r4   rU   ÚdeviceÚrope_init_fnrT   r6   s        €r7   r.   zDogeRotaryEmbedding.__init__M   sÊ   ø€ Ý‰Œ×ÒÑÔÐØ"(Ô"@ˆÔØ$*Ô$BˆÔ!àˆŒàœÔ4°[ÔAˆŒØ!%Ô!EˆØŒ>˜YÒ&Ð&Ý.¨t¬~Ô>ˆLØ+7¨<¸¼ÀVÑ+LÔ+LÑ(ˆ�$Ô(à×Ò˜Z¨¸eÐÑDÔDÐDØ×ÒÐ0°(·.².Ñ2BÔ2BÈuÐÑUÔUÐUÐUÐUr8   rc   ztorch.deviceÚseq_lenr+   ztorch.Tensorc                 óü   — | j         d         }t          | dd¦  «        p| j        | j        z  }d}d|t	          j        d|dt          j        ¬¦  «                             |t          j        ¬¦  «        |z  z  z  }||fS )	a¨  
        Computes the inverse frequencies according to the original RoPE implementation
        Args:
            config ([`~transformers.PreTrainedConfig`]):
                The model configuration.
            device (`torch.device`):
                The device to use for initialization of the inverse frequencies.
            seq_len (`int`, *optional*):
                The current sequence length. Unused for this type of RoPE.
        Returns:
            Tuple of (`torch.Tensor`, `float`), containing the inverse frequencies for the RoPE embeddings and the
            post-processing scaling factor applied to the computed cos/sin (unused in this type of RoPE).
        Ú
rope_thetaÚhead_dimNç      ð?r   r;   ©r>   ©rc   r>   )	r^   Úgetattrr5   Únum_attention_headsr0   ÚarangeÚint64r?   rN   )rU   rc   re   ÚbaseÚdimÚattention_factorrT   s          r7   r_   z3DogeRotaryEmbedding.compute_default_rope_parameters]   sŒ   € ð& Ô% lÔ3ˆÝ�f˜j¨$Ñ/Ô/Ðc°6Ô3EÈÔIcÑ3cˆàÐð Ø•U”\ ! S¨!µ5´;Ð?Ñ?Ô?×BÒBÈ&ÕX]ÔXcÐBÑdÔdÐgjÑjÑkñ
ˆð Ð)Ð)Ð)r8   c                 óN  — | j         d d d …d f                              ¦   «                              |j        d         dd¦  «                             |j        ¦  «        }|d d …d d d …f                              ¦   «         }t          |j        j        t          ¦  «        r|j        j        dk    r|j        j        nd}t          |d¬¦  «        5  |                     ¦   «         |                     ¦   «         z   
                    dd¦  «        }t          j        ||fd¬	¦  «        }|                     ¦   «         | j        z  }|                     ¦   «         | j        z  }	d d d ¦  «         n# 1 swxY w Y   |                     |j        ¬
¦  «        |	                     |j        ¬
¦  «        fS )Nr   r<   r#   ÚmpsÚcpuF)Údevice_typeÚenabledr;   ©rq   rj   )rT   rN   ÚexpandrI   r?   rc   Ú
isinstanceÚtypeÚstrr   Ú	transposer0   ÚcatÚcosr`   Úsinr>   )
r4   ÚxÚposition_idsÚinv_freq_expandedÚposition_ids_expandedrv   ÚfreqsÚembr   r€   s
             r7   rF   zDogeRotaryEmbedding.forward{   s·  € ð !œM¨$°°°°4¨-Ô8×>Ò>Ñ@Ô@×GÒGÈÔHZÐ[\ÔH]Ð_aÐcdÑeÔe×hÒhÐijÔiqÑrÔrÐØ ,¨Q¨Q¨Q°°a°a°a¨ZÔ 8× >Ò >Ñ @Ô @Ðå'1°!´(´-ÅÑ'EÔ'EÐkÈ!Ì(Ì-Ð[`ÒJ`ÐJ`�a”h”m�mÐfkˆÝ¨¸UÐCÑCÔCð 	5ð 	5Ø&×,Ò,Ñ.Ô.Ð1F×1LÒ1LÑ1NÔ1NÑN×YÒYÐZ[Ð]^Ñ_Ô_ˆEÝ”)˜U E˜N°Ð3Ñ3Ô3ˆCØ—'’'‘)”)˜dÔ4Ñ4ˆCØ—'’'‘)”)˜dÔ4Ñ4ˆCð		5ð 	5ð 	5ñ 	5ô 	5ð 	5ð 	5ð 	5ð 	5ð 	5ð 	5øøøð 	5ð 	5ð 	5ð 	5ð �vŠv˜AœGˆvÑ$Ô$ c§f¢f°1´7 fÑ&;Ô&;Ð;Ð;s   ÃBE&Å&E*Å-E*©N)NNN)rK   rL   rM   r0   rO   Ú__annotations__r$   r.   Ústaticmethodr   ÚintrH   rN   r_   Úno_gradr   rF   rP   rQ   s   @r7   rS   rS   J   sù   ø€ € € € € € ØŒlÐÐÑðVð V˜zð Vð Vð Vð Vð Vð Vð  à$(Ø+/Ø"ð*ð *Ø˜TÑ!ð*à˜Ô(ð*ð �t‘ð*ð 
ˆ~˜uÐ$Ô	%ð	*ð *ð *ñ „\ð*ð: €U„]�_„_Øð<ð <ñ Ôñ „_ð<ð <ð <ð <ð <r8   rS   c                 óœ   — | dd| j         d         dz  …f         }| d| j         d         dz  d…f         }t          j        | |fd¬¦  «        S )z*Rotates half the hidden dims of the input..Nr<   r;   rx   )rI   r0   r~   )r�   Úx1Úx2s      r7   Úrotate_halfr�   ‹   s]   € à	
ˆ3Ð"�!”'˜"”+ Ñ"Ð"Ð"Ô	#€BØ	
ˆ3�”˜”˜qÑ Ð"Ð"Ð"Ô	#€BÝŒ9�r�c˜2�Y BÐ'Ñ'Ô'Ð'r8   Úrotary_pos_embc                 ó¾   — |                      |¦  «        }|                      |¦  «        }| |z  t          | ¦  «        |z  z   }||z  t          |¦  «        |z  z   }||fS )a…  Applies Rotary Position Embedding to the query and key tensors.

    Args:
        q (`torch.Tensor`): The query tensor.
        k (`torch.Tensor`): The key tensor.
        cos (`torch.Tensor`): The cosine part of the rotary embedding.
        sin (`torch.Tensor`): The sine part of the rotary embedding.
        unsqueeze_dim (`int`, *optional*, defaults to 1):
            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
    Returns:
        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
    )Ú	unsqueezer�   )ÚqÚkr   r€   Úunsqueeze_dimÚq_embedÚk_embeds          r7   Úapply_rotary_pos_embr˜   ’   sc   € ð& �-Š-˜Ñ
&Ô
&€CØ
�-Š-˜Ñ
&Ô
&€CØ�3‰w�; q™>œ>¨CÑ/Ñ0€GØ�3‰w�; q™>œ>¨CÑ/Ñ0€GØ�GÐÐr8   r9   Ún_repr+   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)rI   ry   Úreshape)r9   r™   ÚbatchÚnum_key_value_headsÚslenrh   s         r7   Ú	repeat_kvrŸ   ¬   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Tr8   ç        ÚmoduleÚ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   r<   )rq   r>   ©ÚpÚtrainingr#   )rŸ   Únum_key_value_groupsr0   Úmatmulr}   r   Ú
functionalÚsoftmaxr@   r?   r>   r§   r¬   Ú
contiguous)r¡   r¢   r£   r¤   r¥   r¦   r§   r¨   Ú
key_statesÚvalue_statesÚattn_weightsÚattn_outputs               r7   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$Ð$r8   r%   Úsoftcapc           
      óP  ‡‡— d }d Št          |t          ¦  «        r|}n|Š‰�‰d d …d d …d d …d |j        d         …f         Šˆˆfd„}	t          ||||	|d|d¬¦  «        \  }
}|                     |j        ¦  «        }|
                     dd¦  «                             ¦   «         }
|
|fS )Néþÿÿÿc                 ó~   •— ‰�‰t          j        | ‰z  ¦  «        z  } ‰�| ‰|         |         |         |         z   } | S r‡   )r0   Útanh)ÚscoreÚ	batch_idxÚhead_idxÚq_idxÚkv_idxÚcausal_maskr·   s        €€r7   Ú	score_modz)flex_attention_forward.<locals>.score_modå   sJ   ø€ ØÐØ�eœj¨°©Ñ9Ô9Ñ9ˆEØÐ"Ø˜K¨	Ô2°8Ô<¸UÔCÀFÔKÑKˆEØˆr8   T)rÂ   Ú
block_maskÚ
enable_gqaÚscaleÚ
return_lser#   r;   )rz   r%   rI   r   r?   r>   r}   r±   )r¡   r¢   r£   r¤   r¥   r¦   r·   r¨   rÃ   rÂ   rµ   Úattention_weightsrÁ   s         `     @r7   Úflex_attention_forwardrÈ   Ñ   s÷   øø€ ð €JØ€KÝ�.¥)Ñ,Ô,ð %Ø#ˆ
ˆ
à$ˆàÐØ! ! ! ! Q Q Q¨¨¨¨?¨S¬Y°r¬]¨?Ð":Ô;ˆðð ð ð ð ð õ &EØØØØØØØð ð&ñ &ô &Ñ"€KÐ"ð *×,Ò,¨U¬[Ñ9Ô9ÐØ×'Ò'¨¨1Ñ-Ô-×8Ò8Ñ:Ô:€KàÐ)Ð)Ð)r8   Údoge_flex_attentionc                   ó   ‡ — e Zd Zddededz  fˆ fd„Z	 	 ddej        deej        ej        f         dej        dz  de	dz  d	eej        ej        dz  eej                 dz  f         f
d
„Z
	 	 ddej        dej        dedej        dz  fd„Zˆ xZS )ÚDogeAttentionNrU   Ú	layer_idxc                 óþ  •— t          ¦   «                              ¦   «          || _        || _        t	          |d|j        |j        z  ¦  «        | _        |j        |j        z  | _	        | j        dz  | _
        |j        | _        |j        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        |j        |j        | j        z  |j        ¬¦  «        | _        t          j        t)          j        |j        ¦  «        ¦  «        | _        t          j        |j        | j        z  |j        |j        ¬¦  «        | _        t          j        |j        | j        z  |j        |j        ¬¦  «        | _        t3          | j        |j        ¬¦  «        | _        t3          | j        |j        ¬¦  «        | _        d S )Nrh   g      à¿©Úbias©r*   )r-   r.   rU   rÌ   rl   r5   rm   rh   r�   r­   r¦   Úattention_dropoutÚkeep_window_sizer   ÚLinearÚattention_biasÚq_projÚk_projÚv_projr/   r0   ÚzerosÚAÚdt_projÚo_projr(   Úrms_norm_epsÚq_normÚk_norm©r4   rU   rÌ   r6   s      €r7   r.   zDogeAttention.__init__  sÊ  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 &Ô 7ˆÔ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 ”�eœk¨&Ô*DÑEÔEÑFÔFˆŒÝ”yØÔ&¨¬Ñ6¸Ô8RÐY_ÔYnð
ñ 
ô 
ˆŒõ ”iØÔ&¨¬Ñ6¸Ô8JÐQWÔQfð
ñ 
ô 
ˆŒõ " $¤-°VÔ5HÐIÑIÔIˆŒÝ! $¤-°VÔ5HÐIÑIÔIˆŒˆˆr8   r9   Úposition_embeddingsr¥   Úpast_key_valuesr+   c                 ó  — |j         d d…         }g |¢d‘| j        ‘R }|                      |                      |¦  «                             |¦  «        ¦  «                             dd¦  «        }|                      |                      |¦  «                             |¦  «        ¦  «                             dd¦  «        }	|                      |¦  «                             |¦  «                             dd¦  «        }
|\  }}t          ||	||¦  «        \  }}	|�| 
                    |	|
| j        ¦  «        \  }	}
|                      |
                     dd¦  «                             |
j         d         |
j         d         d¦  «        ¦  «        }t          j        | j        t#          j        |¦  «        z  ¦  «                             dd¦  «        }|                      ||| j        |¬¦  «        }t+          || j        ¦  «        }t.                               | j        j        t6          ¦  «        } || ||	|
f|| j        sdn| j        | j        dœ|¤Ž\  }} |j        g |¢d‘R Ž                      ¦   «         }|                       |¦  «        }||fS )	Nr<   r#   r;   r   r¹   )r9   Ú	dt_statesrÒ   r¥   r    )r¥   r§   r¦   )!rI   rh   rÝ   rÕ   Úviewr}   rÞ   rÖ   r×   r˜   ÚupdaterÌ   rÚ   r›   r0   ÚexprÙ   ÚFÚsoftplusÚprepare_dynamic_maskrÒ   rŸ   r­   ÚALL_ATTENTION_FUNCTIONSÚget_interfacerU   Ú_attn_implementationr¶   r¬   rÑ   r¦   r±   rÛ   )r4   r9   rà   r¥   rá   r¨   Úinput_shapeÚhidden_shapeÚquery_statesr²   r³   r   r€   rã   Ú	attn_maskÚattention_interfacerµ   r´   s                     r7   rF   zDogeAttention.forward"  s˜  € ð $Ô)¨#¨2¨#Ô.ˆØ8˜Ð8 bÐ8¨$¬-Ð8Ð8ˆà—{’{ 4§;¢;¨}Ñ#=Ô#=×#BÒ#BÀ<Ñ#PÔ#PÑQÔQ×[Ò[Ð\]Ð_`ÑaÔaˆØ—[’[ §¢¨]Ñ!;Ô!;×!@Ò!@ÀÑ!NÔ!NÑOÔO×YÒYÐZ[Ð]^Ñ_Ô_ˆ
Ø—{’{ =Ñ1Ô1×6Ò6°|ÑDÔD×NÒNÈqÐRSÑTÔTˆà&‰ˆˆSÝ#7¸ÀjÐRUÐWZÑ#[Ô#[Ñ ˆ�jàÐ&Ø'6×'=Ò'=¸jÈ,ÐX\ÔXfÑ'gÔ'gÑ$ˆJ˜ð —L’LØ×"Ò" 1 aÑ(Ô(×0Ò0°Ô1CÀAÔ1FÈÔHZÐ[]ÔH^Ð`bÑcÔcñ
ô 
ˆ	õ ”I˜dœf¥q¤z°)Ñ'<Ô'<Ñ<Ñ=Ô=×GÒGÈÈBÑOÔOˆ	Ø×-Ò-Ø'ØØ!Ô2Ø)ð	 .ñ 
ô 
ˆ	õ ˜i¨Ô)BÑCÔCˆ	å(?×(MÒ(MØŒKÔ,Õ.Eñ)
ô )
Ðð %8Ð$7ØØØØð		%
ð
 %Ø#œ}ÐH�C�C°$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(r8   é   rã   rÒ   c           	      óâ  — t          j        |j        ¦  «        j        }|j        }|dd…dd…ddd…f                              dd|j        d         d¦  «        }|�˜t          |t          ¦  «        sƒ|j        t           j        k    r7|j        }t          j	        |t          j
        d|j        |¬¦  «        |¦  «        }|                     |dd…dd…dd…d|j        d         …f         dk    |¦  «        }|j        d         |k    rkt          j        |||j        ¬¦  «        }t          j        ||ddd	¬
¦  «        j        }	|                     d|	d¦  «        }|                     |dk    |¦  «        }|S )a8  
        The core idea of DMA is to calculate the dynamic attention mask to mask the tokens that should be masked, so as to form sparse attention.

        Combine `dt_states` with `attention_mask` to generate the final `attn_mask`.

        Args:
            hidden_states (`torch.Tensor`): The input hidden_states, used to determine the minimum value of the current input precision.
            dt_states (`torch.Tensor`): dt_states of shape `(batch_size, num_heads, key_sequence_length)`.
            keep_window_size (`int`): The window size of tokens that are not dynamically masked, and dynamic masking is only performed when the sequence length exceeds this value.
            attention_mask (`torch.Tensor`, *optional*): attention mask of shape `(batch_size, 1, query_sequence_length, key_sequence_length)`.
        Nr<   r#   r    rk   r   ©r>   rc   TF)rq   ÚlargestÚsortedri   )r0   Úfinfor>   Úminry   rI   rz   r%   ÚboolÚwhereÚtensorrc   Úmasked_fillÚ
zeros_likeÚtopkÚindicesÚscatter)
r4   r9   rã   rÒ   r¥   Ú	min_dtyper>   rð   Úactive_maskÚtopk_indicess
             r7   ré   z"DogeAttention.prepare_dynamic_maskW  s‡  € õ$ ”K Ô 3Ñ4Ô4Ô8ˆ	ØÔ#ˆØ˜a˜a˜a    D¨!¨!¨!˜mÔ,×3Ò3Ø��MÔ'¨Ô*¨Bñ
ô 
ˆ	ð Ð%­j¸ÍÑ.SÔ.SÐ%ØÔ#¥u¤zÒ1Ð1Ø%Ô+�Ý!&¤Ø"¥E¤L°¸^Ô=RÐZ_Ð$`Ñ$`Ô$`Ðbkñ"ô "�ð "×-Ò-¨n¸Q¸Q¸QÀÀÀÀ1À1À1ÐF[È	ÌÐXZÔH[ÐF[Ð=[Ô.\Ð`aÒ.aÐclÑmÔmˆIØŒ?˜2ÔÐ!1Ò1Ð1ÝÔ*¨9¸EÈ)ÔJZÐ[Ñ[Ô[ˆKÝ œ: iÐ1AÀrÐSWÐ`eÐfÑfÔfÔnˆLØ%×-Ò-¨b°,ÀÑDÔDˆKØ!×-Ò-¨k¸SÒ.@À)ÑLÔLˆIØÐr8   r‡   ©NN)rò   N)rK   rL   rM   r$   rŠ   r.   r0   rO   rH   r
   rF   ré   rP   rQ   s   @r7   rË   rË     s>  ø€ € € € € ðJð J˜zð J°c¸D±jð Jð Jð Jð Jð Jð JðD /3Ø(,ð3)ð 3)à”|ð3)ð # 5¤<°´Ð#=Ô>ð3)ð œ tÑ+ð	3)ð
  ™ð3)ð 
ˆuŒ|˜Uœ\¨DÑ0°%¸¼Ô2EÈÑ2LÐLÔ	Mð3)ð 3)ð 3)ð 3)ðr !%Ø.2ð#ð #à”|ð#ð ”<ð#ð ð	#ð
 œ tÑ+ð#ð #ð #ð #ð #ð #ð #ð #r8   rË   c                   ó$   ‡ — e Zd Zˆ fd„Zd„ Zˆ xZS )ÚDogeMLPc                 ó¶  •— t          ¦   «                              ¦   «          || _        |j        | _        |j        | _        t          j        | j        | j        |j        ¬¦  «        | _        t          j        | j        | j        |j        ¬¦  «        | _	        t          j        | j        | j        |j        ¬¦  «        | _
        t          |j                 | _        d S )NrÎ   )r-   r.   rU   r5   Úintermediate_sizer   rÓ   Úmlp_biasÚ	gate_projÚup_projÚ	down_projr	   Ú
hidden_actÚact_fn©r4   rU   r6   s     €r7   r.   zDogeMLP.__init__~  s¯   ø€ Ý‰Œ×ÒÑÔÐØˆŒØ!Ô-ˆÔØ!'Ô!9ˆÔÝœ 4Ô#3°TÔ5KÐRXÔRaÐbÑbÔbˆŒÝ”y Ô!1°4Ô3IÐPVÔP_Ð`Ñ`Ô`ˆŒÝœ 4Ô#9¸4Ô;KÐRXÔRaÐbÑbÔbˆŒÝ˜VÔ.Ô/ˆŒˆˆr8   c                 ó¨   — |                       |                      |                      |¦  «        ¦  «        |                      |¦  «        z  ¦  «        }|S r‡   )r  r  r
  r  )r4   r�   r  s      r7   rF   zDogeMLP.forwardˆ  sA   € Ø—N’N 4§;¢;¨t¯~ª~¸aÑ/@Ô/@Ñ#AÔ#AÀDÇLÂLÐQRÁOÄOÑ#SÑTÔTˆ	ØÐr8   )rK   rL   rM   r.   rF   rP   rQ   s   @r7   r  r  }  sG   ø€ € € € € ð0ð 0ð 0ð 0ð 0ðð ð ð ð ð ð r8   r  c                   óH   ‡ — e Zd Zdefˆ fd„Zdej        dej        fd„Zˆ xZS )Ú	DogeCDMoErU   c                 ó2  •— t          ¦   «                              ¦   «          |j        | _        |j        | _        t          |j                 | _        |j        | _        t          j	        t          j
        | j        ¦  «        ¦  «        | _        |j        | _        |j        | _        t          j        | j        | j        |j        ¬¦  «        | _        t          j        | j        | j        |j        ¬¦  «        | _        t          j        | j        | j        |j        ¬¦  «        | _        t          j        | j        | j        dz  d¬¦  «        | _        t          j        | j        | j        ¦  «        | _        t          j        | j        | j        ¦  «        | _        d S )NrÎ   r;   F)r-   r.   r5   r  r	   r  r  Únum_expertsÚmathÚfloorÚsqrtÚnum_keysÚnum_experts_per_tokÚtop_kÚnorm_topk_probr   rÓ   r	  r
  r  r  Úrouter_gateÚ	EmbeddingÚ
down_embedÚup_embedr  s     €r7   r.   zDogeCDMoE.__init__Ž  sA  ø€ Ý‰Œ×ÒÑÔÐØ!Ô-ˆÔØ!'Ô!9ˆÔÝ˜VÔ.Ô/ˆŒà!Ô-ˆÔÝœ
¥4¤9¨TÔ-=Ñ#>Ô#>Ñ?Ô?ˆŒØÔ/ˆŒ
Ø$Ô3ˆÔõ œ 4Ô#3°TÔ5KÐRXÔRaÐbÑbÔbˆŒÝ”y Ô!1°4Ô3IÐPVÔP_Ð`Ñ`Ô`ˆŒÝœ 4Ô#9¸4Ô;KÐRXÔRaÐbÑbÔbˆŒõ œ9 TÔ%5°t´}ÀqÑ7HÈuÐUÑUÔUˆÔõ œ, tÔ'7¸Ô9IÑJÔJˆŒÝœ TÔ%5°tÔ7GÑHÔHˆŒˆˆr8   r9   r+   c                 ó  — |j         \  }}}|                      |¦  «                             d||z  d¦  «        }|                     | j        d¬¦  «        \  \  }}\  }	}
|                     d¦  «        |                     d¦  «        z   }|	                     d¦  «        | j        z  |
                     d¦  «        z   } |j        g |j         d d…         ¢d‘R Ž } |j        g |j         d d…         ¢d‘R Ž }|                     | j        d¬¦  «        \  }}|                     d|¦  «        }t          j	        |d¬¦  «        }| j
        r||                     dd¬¦  «        z  }|                      |¦  «        }|                      |¦  «        }t          j        ||                     ||z  dd¦  «        ¦  «                             ||z  d¦  «        }|                      |¦  «        |z  }t          j        |                     ||z  dd¦  «        |¦  «                             ||d¦  «        }|                      |                      |                      |¦  «        ¦  «        |                      |¦  «        z  ¦  «        }||z   }||fS )Nr;   r<   rx   r¹   T)rq   r=   r#   )rI   r  rä   rþ   r  r’   r  Úgatherrç   r°   r  Úsumr  r  r0   r®   r  r  r
  r  )r4   r9   r¨   Úbszre   Ú_Úrouter_logitsÚscores_xÚscores_yÚ	indices_xÚ	indices_yÚ
all_scoresÚall_indicesÚscoresÚposition_indicesrÿ   Úrouting_weightsr  r  Úexperts_weightsÚexperts_statess                        r7   rF   zDogeCDMoE.forward¥  sŠ  € ð
 (Ô-‰ˆˆW�að ×(Ò(¨Ñ7Ô7×<Ò<¸QÀÀgÁÈrÑRÔRˆð 8E×7IÒ7IÈ$Ì-Ð]_Ð7IÑ7`Ô7`Ñ4Ñˆ�8Ñ4˜y¨)Ø×'Ò'¨Ñ+Ô+¨h×.@Ò.@ÀÑ.DÔ.DÑDˆ
Ø×)Ò)¨"Ñ-Ô-°´Ñ=À	×@SÒ@SÐTVÑ@WÔ@WÑWˆØ$�Z”_Ð@ jÔ&6°s¸°sÔ&;Ð@¸RÐ@Ð@Ð@ˆ
Ø&�kÔ&ÐC¨Ô(9¸#¸2¸#Ô(>ÐCÀÐCÐCÐCˆØ#-§?¢?°4´:À2 ?Ñ#FÔ#FÑ ˆÐ Ø×$Ò$ RÐ)9Ñ:Ô:ˆÝœ) F°Ð3Ñ3Ô3ˆØÔð 	IØ˜×2Ò2°rÀ4Ð2ÑHÔHÑHˆOð —_’_ WÑ-Ô-ˆ
Ø—=’= Ñ)Ô)ˆÝœ, z°=×3EÒ3EÀcÈGÁmÐUWÐYZÑ3[Ô3[Ñ\Ô\×aÒaÐbeÐhoÑboÐqsÑtÔtˆØŸ+š+ oÑ6Ô6¸ÑHˆÝœ o×&:Ò&:¸3À¹=È!ÈRÑ&PÔ&PÐRZÑ[Ô[×`Ò`ÐadÐfmÐoqÑrÔrˆØŸš t§{¢{°4·>²>À-Ñ3PÔ3PÑ'QÔ'QÐTX×T`ÒT`ÐanÑToÔToÑ'oÑpÔpˆØ%¨Ñ6ˆØ˜mÐ+Ð+r8   )	rK   rL   rM   r$   r.   r0   rO   rF   rP   rQ   s   @r7   r  r  �  su   ø€ € € € € ðI˜zð Ið Ið Ið Ið Ið Ið.,à”|ð,ð 
Œð	,ð ,ð ,ð ,ð ,ð ,ð ,ð ,r8   r  c                   ó  ‡ — e Zd Zddededz  fˆ fd„Z	 	 	 	 	 ddej        deej        ej        f         dz  dej        dz  d	ej	        dz  d
e
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 )ÚDogeDecoderLayerNrU   rÌ   c                 ó   •— t          ¦   «                              ¦   «          |j        | _        t          |j        |j        ¬¦  «        | _        t          ||¬¦  «        | _        t          j
        t          j        |j        ¦  «        ¦  «        | _        t          |j        |j        ¬¦  «        | _        |j        st!          |¦  «        nt#          |¦  «        | _        t          j
        t          j        |j        ¦  «        ¦  «        | _        d S )NrÐ   )rU   rÌ   )r-   r.   Úhidden_dropoutr(   r5   rÜ   Úinput_layernormrË   Ú	self_attnr   r/   r0   r1   Úinput_residualÚpost_attention_layernormÚis_moer  r  ÚmlpÚpost_attention_residualrß   s      €r7   r.   zDogeDecoderLayer.__init__Ç  sÒ   ø€ Ý‰Œ×ÒÑÔÐØ$Ô3ˆÔå*¨6Ô+=À6ÔCVÐWÑWÔWˆÔÝ&¨fÀ	ÐJÑJÔJˆŒÝ œl­5¬:°fÔ6HÑ+IÔ+IÑJÔJˆÔå(3°FÔ4FÈFÔL_Ð(`Ñ(`Ô(`ˆÔ%Ø*0¬-ÐN•7˜6‘?”?�?½YÀvÑ=NÔ=NˆŒÝ')¤|µE´J¸vÔ?QÑ4RÔ4RÑ'SÔ'SˆÔ$Ð$Ð$r8   Fr9   rà   r¥   r‚   rá   Ú	use_cacher¨   r+   c           
      ór  — |}|                       |¦  «        } | j        d||||||dœ|¤Ž\  }}	t          j        || j        | j        ¬¦  «        }| j        |z  |z   }|}|                      |¦  «        }|                      |¦  «        }t          j        || j        | j        ¬¦  «        }| j	        |z  |z   }|S )N)r9   rà   r¥   r‚   rá   r<  rª   © )
r5  r6  rç   r§   r4  r¬   r7  r8  r:  r;  )
r4   r9   rà   r¥   r‚   rá   r<  r¨   ÚresidualÚself_attn_weightss
             r7   rF   zDogeDecoderLayer.forwardÓ  sì   € ð !ˆØ×,Ò,¨]Ñ;Ô;ˆØ+9¨4¬>ð ,
Ø'Ø 3Ø)Ø%Ø+Øð,
ð ,
ð ð,
ð ,
Ñ(ˆÐ(õ œ	 -°4Ô3FÐQUÔQ^Ð_Ñ_Ô_ˆØÔ+¨hÑ6¸ÑFˆð !ˆØ×5Ò5°mÑDÔDˆØŸš Ñ/Ô/ˆÝœ	 -°4Ô3FÐQUÔQ^Ð_Ñ_Ô_ˆØÔ4°xÑ?À-ÑOˆàÐr8   r‡   )NNNNF)rK   rL   rM   r$   rŠ   r.   r0   rO   rH   Ú
LongTensorr
   rù   r   r   ÚFloatTensorrF   rP   rQ   s   @r7   r2  r2  Æ  s,  ø€ € € € € ð
Tð 
T˜zð 
T°c¸D±jð 
Tð 
Tð 
Tð 
Tð 
Tð 
Tð IMØ.2Ø04Ø(,Ø!&ð ð  à”|ð ð # 5¤<°´Ð#=Ô>ÀÑEð ð œ tÑ+ð	 ð
 Ô&¨Ñ-ð ð  ™ð ð ˜$‘;ð ð Ð+Ô,ð ð 
ˆuÔ  %¨Ô(9¸5Ô;LÐ(LÔ"MÐPTÑ"TÐTÔ	Uð ð  ð  ð  ð  ð  ð  ð  r8   r2  c                   óž   ‡ — e Zd ZU eed<   dZdZdgZdgZdZ	dZ
dZdZdZ eed¬¦  «        eed	œZ ej        ¦   «         ˆ fd
„¦   «         Zˆ xZS )ÚDogePreTrainedModelrU   ÚmodelTr2  rá   Fr#   )Úindex)r%  r9   Ú
attentionsc                 ó¤  •— t          ¦   «                              |¦  «         t          |t          ¦  «        r-t	          |d¦  «        rt          j        |j        ¦  «         dS dS t          |t          ¦  «        rTt	          |d¦  «        rt          j	        |j
        ¦  «         t	          |d¦  «        rt          j	        |j        ¦  «         dS dS dS )zInitialize the weightsrÙ   r7  r;  N)r-   Ú_init_weightsrz   rË   ÚhasattrÚinitÚzeros_rÙ   r2  Úones_r7  r;  )r4   r¡   r6   s     €r7   rI  z!DogePreTrainedModel._init_weights  sÜ   ø€ õ 	‰Œ×Ò˜fÑ%Ô%Ð%Ý�f�mÑ,Ô,ð 	;Ý�v˜sÑ#Ô#ð &Ý”˜FœHÑ%Ô%Ð%Ð%Ð%ð&ð &å˜Õ 0Ñ1Ô1ð 	;Ý�vÐ/Ñ0Ô0ð 2Ý”
˜6Ô0Ñ1Ô1Ð1Ý�vÐ8Ñ9Ô9ð ;Ý”
˜6Ô9Ñ:Ô:Ð:Ð:Ð:ð		;ð 	;ð;ð ;r8   )rK   rL   rM   r$   rˆ   Úbase_model_prefixÚsupports_gradient_checkpointingÚ_no_split_modulesÚ_skip_keys_device_placementÚ_supports_flash_attnÚ_supports_sdpaÚ_supports_flex_attnÚ_can_compile_fullgraphÚ_supports_attention_backendr!   r  r2  rË   Ú_can_record_outputsr0   r‹   rI  rP   rQ   s   @r7   rD  rD  ö  sº   ø€ € € € € € àÐÐÑØÐØ&*Ð#Ø+Ð,ÐØ#4Ð"5ÐØ ÐØ€NØÐØ"ÐØ"&Ðà'˜¨	¸Ð;Ñ;Ô;Ø)Ø#ðð Ðð €U„]�_„_ð
;ð 
;ð 
;ð 
;ñ „_ð
;ð 
;ð 
;ð 
;ð 
;r8   rD  c                   óâ   ‡ — e Zd Zdefˆ fd„Zeee	 	 	 	 	 	 ddej	        dz  dej
        dz  dej	        dz  dedz  dej        dz  d	edz  d
ee         defd„¦   «         ¦   «         ¦   «         Zˆ xZS )Ú	DogeModelrU   c                 óÞ  •‡— t          ¦   «                              ‰¦  «         ‰j        | _        ‰j        | _        t          j        ‰j        ‰j        | j        ¦  «        | _        t          j	        ˆfd„t          ‰j        ¦  «        D ¦   «         ¦  «        | _        t          ‰j        ‰j        ¬¦  «        | _        t!          ‰¬¦  «        | _        d| _        |                      ¦   «          d S )Nc                 ó0   •— g | ]}t          ‰|¦  «        ‘ŒS r>  )r2  )Ú.0rÌ   rU   s     €r7   ú
<listcomp>z&DogeModel.__init__.<locals>.<listcomp>  s$   ø€ ÐbÐbÐb°YÕ˜f iÑ0Ô0ÐbÐbÐbr8   rÐ   ©rU   F)r-   r.   Úpad_token_idÚpadding_idxÚ
vocab_sizer   r  r5   Úembed_tokensÚ
ModuleListÚrangeÚnum_hidden_layersÚlayersr(   rÜ   ÚnormrS   Ú
rotary_embÚgradient_checkpointingÚ	post_initr  s    `€r7   r.   zDogeModel.__init__  sÔ   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”mØbÐbÐbÐbÅ%ÈÔH`ÑBaÔBaÐbÑbÔbñ
ô 
ˆŒõ   Ô 2¸Ô8KÐLÑLÔLˆŒ	Ý-°VÐ<Ñ<Ô<ˆŒØ&+ˆÔ#ð 	�ŠÑÔÐÐÐr8   NÚ	input_idsr¥   r‚   rá   Úinputs_embedsr<  r¨   r+   c           
      ót  — |d u |d uz  rt          d¦  «        ‚|r|€t          | j        ¬¦  «        }|€|                      |¦  «        }|€V|�|                     ¦   «         nd}t          j        |j        d         |j        ¬¦  «        |z   }| 	                    d¦  «        }| j        j
        €t          nt          }	 |	| j        ||||¬¦  «        }
|}|                      ||¬¦  «        }| j        d | j        j        …         D ]} ||f|
||||dœ|¤Ž}Œ|                      |¦  «        }t#          ||¬	¦  «        S )
Nz:You must specify exactly one of input_ids or inputs_embedsr^  r   r#   )rc   )rU   rl  r¥   rá   r‚   )r‚   )r¥   r‚   rá   r<  rà   )Úlast_hidden_staterá   )Ú
ValueErrorr   rU   rb  Úget_seq_lengthr0   rn   rI   rc   r’   Úsliding_windowr   r   rh  rf  re  rg  r   )r4   rk  r¥   r‚   rá   rl  r<  r¨   Úpast_seen_tokensÚmask_functionrÁ   r9   rà   Údecoder_layers                 r7   rF   zDogeModel.forward(  s¢  € ð ˜Ð -°tÐ";Ñ<ð 	[ÝÐYÑZÔZÐZàð 	?˜Ð0Ý*°$´+Ð>Ñ>Ô>ˆOàÐ Ø ×-Ò-¨iÑ8Ô8ˆMàÐØCRÐC^˜×=Ò=Ñ?Ô?Ð?ÐdeÐÝ œ<¨Ô(;¸AÔ(>À}ÔG[Ð\Ñ\Ô\Ð_oÑoˆLØ'×1Ò1°!Ñ4Ô4ˆLà.2¬kÔ.HÐ.PÕ*Ð*ÕVwˆØ#�mØ”;Ø'Ø)Ø+Ø%ð
ñ 
ô 
ˆð &ˆØ"Ÿošo¨mÈ,˜oÑWÔWÐà!œ[Ð)H¨4¬;Ô+HÐ)HÔIð 		ð 		ˆMØ)˜MØðà*Ø)Ø /Ø#Ø$7ðð ð ðð ˆMˆMð Ÿ	š	 -Ñ0Ô0ˆå%Ø+Ø+ð
ñ 
ô 
ð 	
r8   )NNNNNN)rK   rL   rM   r$   r.   r    r"   r   r0   rA  rO   r
   rB  rù   r   r   r   rF   rP   rQ   s   @r7   rY  rY    s  ø€ € € € € ð˜zð ð ð ð ð ð ð   ØØð .2Ø.2Ø04Ø(,Ø26Ø!%ð4
ð 4
àÔ# dÑ*ð4
ð œ tÑ+ð4
ð Ô&¨Ñ-ð	4
ð
  ™ð4
ð Ô(¨4Ñ/ð4
ð ˜$‘;ð4
ð Ð+Ô,ð4
ð 
 ð4
ð 4
ð 4
ñ „^ñ „_ñ  Ôð4
ð 4
ð 4
ð 4
ð 4
r8   rY  r;   Úgate_logitsr  r  r  c                 óT  — | �t          | t          ¦  «        sdS | d         j        }| d         j        }g }g }| D �]9}	|	                     |¦  «        }	|	                     |d¬¦  «        \  \  }
}\  }}|
                     d¦  «        |                     d¦  «        z   }|                     d¦  «        |z  |                     d¦  «        z   } |j        g |j        dd…         ¢d‘R Ž } |j        g |j        dd…         ¢d‘R Ž }|                     |d¬¦  «        \  }}| 	                    d|¦  «        }t          j        |d¬¦  «        }|                     |¦  «         |                     |¦  «         �Œ;t          j        |d¬¦  «        }t          j        |d¬¦  «        }|€€|                     d¦  «        }t          j        |||¬¦  «        }t          j        |||¬¦  «        }|                     d||¦  «        |j        d         z  }t          j        |d¬¦  «        }�nk|j        \  }}t'          | ¦  «        }|ddd…dd…df                              ||||f¦  «                             d¦  «                             |¦  «        }|                     d¦  «        |                     ¦   «                  }t          j        |||¬¦  «        }t          j        |||¬¦  «        }|                     d||¦  «        t          j        |¦  «        z  }|ddd…dd…df                              ||||f¦  «                             d|¦  «                             |¦  «        }t          j        ||z  d¬¦  «        t          j        |d¬¦  «        z  }t          j        ||z  ¦  «        }||z  S )aø  
    Computes auxiliary load balancing loss as in Switch Transformer - implemented in Pytorch.

    See Switch Transformer (https://huggingface.co/papers/2101.03961) for more details. This function implements the loss
    function presented in equations (4) - (6) of the paper. It aims at penalizing cases where the routing between
    experts is too unbalanced.

    Args:
        gate_logits:
            Logits from the `router_gate`, should be a tuple of model.config.num_hidden_layers tensors of
            shape [2, batch_size * sequence_length, num_keys].
        num_experts:
            Number of experts
        num_keys:
            Number of keys
        top_k:
            The number of experts to route per-token, can be also interpreted as the `top-k` routing
            parameter.
        attention_mask (`torch.Tensor`, *optional*):
            The attention_mask used in forward function
            shape [batch_size X sequence_length] if not None.

    Returns:
        The auxiliary loss.
    Nr   r<   rx   r¹   rô   )rz   rH   r>   rc   r?   rþ   r’   rä   rI   r!  rç   r°   Úappendr0   r~   rØ   Ú	ones_likeÚscatter_add_rB   Úlenry   r›   rù   r"  )ru  r  r  r  r¥   Úcompute_dtypeÚcompute_deviceÚall_expert_indicesÚall_routing_weightsÚlayer_gate_logitsr&  r'  r(  r)  r*  r+  r$  r-  Úexpert_indicesr.  Útokens_per_expertÚpadÚrouter_prob_per_expertÚ
batch_sizeÚsequence_lengthre  Úexpert_attention_maskÚ router_per_expert_attention_maskÚoverall_losss                                r7   Úload_balancing_loss_funcr‰  b  s  € ð@ Ð¥*¨[½%Ñ"@Ô"@ÐØˆqà ”NÔ(€MØ  ”^Ô*€NØÐØÐà(ð 4ñ 4ÐØ-×0Ò0°Ñ@Ô@Ðà7H×7MÒ7MÈhÐ\^Ð7MÑ7_Ô7_Ñ4Ñˆ�8Ñ4˜y¨)à×'Ò'¨Ñ+Ô+¨h×.@Ò.@ÀÑ.DÔ.DÑDˆ
Ø×)Ò)¨"Ñ-Ô-°Ñ8¸9×;NÒ;NÈrÑ;RÔ;RÑRˆØ$�Z”_Ð@ jÔ&6°s¸°sÔ&;Ð@¸RÐ@Ð@Ð@ˆ
Ø&�kÔ&ÐC¨Ô(9¸#¸2¸#Ô(>ÐCÀÐCÐCÐCˆà(Ÿošo¨e¸˜oÑ<Ô<ÑˆÐØ$×+Ò+¨BÐ0@ÑAÔAˆåœ) J°BÐ7Ñ7Ô7ˆà×!Ò! .Ñ1Ô1Ð1Ø×"Ò" ?Ñ3Ô3Ð3Ñ3ÝœÐ#5¸1Ð=Ñ=Ô=ÐÝœ)Ð$7¸QÐ?Ñ?Ô?ÐàÐà/×4Ò4°RÑ8Ô8ÐÝ!œK¨¸=ÐQ_Ð`Ñ`Ô`ÐÝŒoÐ0¸ÈnÐ]Ñ]Ô]ˆØ-×:Ò:¸1Ð>PÐRUÑVÔVÐYkÔYqÐrsÔYtÑtÐõ "'¤Ð,?ÀQÐ!GÑ!GÔ!GÐÑà&4Ô&:Ñ#ˆ
�OÝ Ñ,Ô,Ðð ˜4    A A A tÐ+Ô,ßŠVÐ&¨
°OÀUÐKÑLÔLßŠW�R‰[Œ[ßŠR�ÑÔð	 	ð 0×4Ò4°RÑ8Ô8Ð9N×9SÒ9SÑ9UÔ9UÔVÐõ "œK¨¸=ÐQ_Ð`Ñ`Ô`ÐÝŒoÐ0¸ÈnÐ]Ñ]Ô]ˆØ-×:Ò:¸1Ð>PÐRUÑVÔVÕY^ÔYbØ!ñZ
ô Z
ñ 
Ðð ˜4    A A A tÐ+Ô,ßŠVÐ&¨
°OÀ[ÐQÑRÔRßŠW�R˜Ñ%Ô%ßŠR�ÑÔð	 	)õ "'¤Ð+>ÐAaÑ+aÐghÐ!iÑ!iÔ!iÕlqÔluØ,°!ðm
ñ m
ô m
ñ "
Ðõ ”9Ð.Ð1GÑGÑHÔH€LØ˜+Ñ%Ð%r8   c                   ó$  ‡ — e Zd ZddiZddiZddgdgfiZˆ fd„Z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	j
        dz  dedz  dee	j        z  dedz  dee         defd„¦   «         ¦   «         Zˆ xZS )ÚDogeForCausalLMzlm_head.weightzmodel.embed_tokens.weightÚlm_headÚcolwise_gather_outputr9   Úlogitsc                 óF  •— t          ¦   «                              |¦  «         t          |¦  «        | _        |j        | _        t          j        |j        |j        d¬¦  «        | _        |j	        | _	        |j
        | _
        |j        | _        |                      ¦   «          d S )NFrÎ   )r-   r.   rY  rE  ra  r   rÓ   r5   rŒ  Úrouter_aux_loss_coefr  r  rj  r  s     €r7   r.   zDogeForCausalLM.__init__Ò  s‹   ø€ Ý‰Œ×Ò˜Ñ Ô Ð Ý˜vÑ&Ô&ˆŒ
Ø Ô+ˆŒÝ”y Ô!3°VÔ5FÈUÐSÑSÔSˆŒØ$*Ô$?ˆÔ!Ø!Ô-ˆÔØ#)Ô#=ˆÔ ð 	�ŠÑÔÐÐÐr8   Nr   rk  r¥   r‚   rá   rl  Úlabelsr<  Úlogits_to_keepÚoutput_router_logitsr¨   r+   c
           
      ó^  — |	�|	n| j         j        }	 | j        d||||||dœ|
¤Ž}|j        }t	          |t
          ¦  «        rt          | d¦  «        n|}|                      |dd…|dd…f         ¦  «        }d}|� | j        ||| j	        fi |
¤Ž}d}|	rrt          |j        | j        t          j        t          j        | j        ¦  «        ¦  «        | j        |¦  «        }|�%|| j        |                     |j        ¦  «        z  z  }t)          ||||j        |j        |j        |j        ¬¦  «        S )ah  
        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.

        Example:

        ```python
        >>> from transformers import AutoTokenizer, DogeForCausalLM

        >>> model = DogeForCausalLM.from_pretrained("SmallDoge/Doge-320M")
        >>> tokenizer = AutoTokenizer.from_pretrained("SmallDoge/Doge-320M")

        >>> prompt = "Hey, are you conscious? Can you talk to me?"
        >>> inputs = tokenizer(prompt, return_tensors="pt")

        >>> # Generate
        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
        ```N)rk  r¥   r‚   rá   rl  r<  )ÚlossÚaux_lossrŽ  rá   r9   rG  r%  r>  )rU   r“  rE  rn  rz   rŠ   ÚslicerŒ  Úloss_functionra  r‰  r%  r  r  r  r  r  r�  r?   rc   r   rá   r9   rG  )r4   rk  r¥   r‚   rá   rl  r‘  r<  r’  r“  r¨   Úoutputsr9   Úslice_indicesrŽ  r•  r–  s                    r7   rF   zDogeForCausalLM.forwardÞ  s…  € ðL %9Ð$DÐ Ð È$Ì+ÔJjð 	ð
 +5¨$¬*ð +
ØØ)Ø%Ø+Ø'Øð+
ð +
ð ð+
ð +
ˆð  Ô1ˆå8BÀ>ÕSVÑ8WÔ8WÐk�˜~˜o¨tÑ4Ô4Ð4Ð]kˆØ—’˜m¨A¨A¨A¨}¸a¸a¸aÐ,?Ô@ÑAÔAˆàˆØÐØ%�4Ô% f¨f°d´oÐPÐPÈÐPÐPˆDàˆØð 		MÝ/ØÔ%ØÔ Ý”
�4œ9 TÔ%5Ñ6Ô6Ñ7Ô7ØÔ(Øñô ˆHð Ð!Ø˜Ô1°H·K²KÀÄÑ4LÔ4LÑLÑL�å(ØØØØ#Ô3Ø!Ô/ØÔ)Ø!Ô/ð
ñ 
ô 
ð 	
r8   )	NNNNNNNr   N)rK   rL   rM   Ú_tied_weights_keysÚ_tp_planÚ_pp_planr.   r   r   r0   rA  rO   r
   rB  rù   rŠ   r   r   r   rF   rP   rQ   s   @r7   r‹  r‹  Ì  sp  ø€ € € € € à*Ð,GÐHÐØÐ2Ð3€HØ˜_Ð-°¨zÐ:Ð;€Hð
ð 
ð 
ð 
ð 
ð Øð .2Ø.2Ø04Ø(,Ø26Ø*.Ø!%Ø-.Ø,0ðO
ð O
àÔ# dÑ*ðO
ð œ tÑ+ðO
ð Ô&¨Ñ-ð	O
ð
  ™ðO
ð Ô(¨4Ñ/ðO
ð Ô  4Ñ'ðO
ð ˜$‘;ðO
ð ˜eœlÑ*ðO
ð # T™kðO
ð Ð+Ô,ðO
ð 
#ðO
ð O
ð O
ñ „^ñ ÔðO
ð O
ð O
ð O
ð O
r8   r‹  c                   ó   — e Zd ZdS )ÚDogeForSequenceClassificationN)rK   rL   rM   r>  r8   r7   rŸ  rŸ  2  s   € € € € € Ø€Dr8   rŸ  )r‹  rY  rD  rŸ  )r#   )r    r  )NNr;   N)Qr  Úcollections.abcr   Útypingr   r   r0   Útorch.nn.functionalr   r¯   rç   Ú r   rK  Úactivationsr	   Úcache_utilsr
   r   Ú
generationr   Úintegrationsr   r   Úintegrations.flex_attentionr   Úmasking_utilsr   r   Úmodeling_layersr   r   Úmodeling_outputsr   r   Úmodeling_rope_utilsr   r   Úmodeling_utilsr   r   Úprocessing_utilsr   Úutilsr   r   r   r   Úutils.genericr   r    Úutils.output_capturingr!   r"   Úconfiguration_doger$   Ú!torch.nn.attention.flex_attentionr%   ÚModuler(   rS   r�   r˜   rO   rŠ   rŸ   rN   r¶   rH   rÈ   rê   rË   r  r  r2  rD  rY  r‰  r‹  rŸ  Ú__all__r>  r8   r7   ú<module>r¶     s`  ðð. €€€Ø $Ð $Ð $Ð $Ð $Ð $Ø "Ð "Ð "Ð "Ð "Ð "Ð "Ð "à €€€Ø Ð Ð Ð Ð Ð Ð Ð Ð Ø Ð Ð Ð Ð Ð à &Ð &Ð &Ð &Ð &Ð &Ø !Ð !Ð !Ð !Ð !Ð !Ø .Ð .Ð .Ð .Ð .Ð .Ð .Ð .Ø )Ð )Ð )Ð )Ð )Ð )Ø QÐ QÐ QÐ QÐ QÐ QÐ QÐ QØ JÐ JÐ JÐ JÐ JÐ JØ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RØ [Ð [Ð [Ð [Ð [Ð [Ð [Ð [Ø QÐ QÐ QÐ QÐ QÐ QÐ QÐ QØ KÐ KÐ KÐ KÐ KÐ KÐ KÐ KØ AÐ AÐ AÐ AÐ AÐ AÐ AÐ AØ &Ð &Ð &Ð &Ð &Ð &Ø gÐ gÐ gÐ gÐ gÐ gÐ gÐ gÐ gÐ gÐ gÐ gØ GÐ GÐ GÐ GÐ GÐ GÐ GÐ GØ EÐ EÐ EÐ EÐ EÐ EÐ EÐ EØ *Ð *Ð *Ð *Ð *Ð *ð  ÐÑ!Ô!ð <Ø;Ð;Ð;Ð;Ð;Ð;ð Ð˜YÑ'Ô'ðJð Jð Jð Jð J�"”)ñ Jô Jñ (Ô'ðJð(><ð ><ð ><ð ><ð ><˜"œ)ñ ><ô ><ð ><ðB(ð (ð (ð ÐÐ*Ñ+Ô+ðð ð ñ ,Ô+ðð2	U˜Uœ\ð 	U°#ð 	U¸%¼,ð 	Uð 	Uð 	Uð 	Uð& ð%ð %ØŒIð%àŒ<ð%ð 
Œð%ð Œ<ð	%ð
 ”L 4Ñ'ð%ð ð%ð ð%ð Ð'Ô(ð%ð %ð %ð %ð> !Ø ð+*ð +*ØŒIð+*àŒ<ð+*ð 
Œð+*ð Œ<ð	+*ð
 ˜%œ,¨Ð3Ô4ð+*ð �T‰\ð+*ð �T‰\ð+*ð ˆ5Œ<˜œÐ%Ô&ð+*ð +*ð +*ð +*ð\ -Ð,Ñ.Ô.Ð Ø1GÐ Ð-Ñ .ðwð wð wð wð w�B”Iñ wô wð wðtð ð ð ð ˆbŒiñ ô ð ð 6,ð 6,ð 6,ð 6,ð 6,�”	ñ 6,ô 6,ð 6,ðr-ð -ð -ð -ð -Ð1ñ -ô -ð -ð` ð;ð ;ð ;ð ;ð ;˜/ñ ;ô ;ñ „ð;ð> ðH
ð H
ð H
ð H
ð H
Ð#ñ H
ô H
ñ „ðH
ðZ #ØØØ*.ðg&ð g&Ø”  e¤lÔ 3Ñ3°dÑ:ðg&à�t‘ðg&ð �D‰jðg&ð ð	g&ð
 ”L 4Ñ'ðg&ð „\�CÑðg&ð g&ð g&ð g&ðT ðb
ð b
ð b
ð b
ð b
Ð)¨?ñ b
ô b
ñ „ðb
ðJ	ð 	ð 	ð 	ð 	Ð$DÐFYñ 	ô 	ð 	ð cÐ
bÐ
b€€€r8   