§
    9²¸jP€  ã                   ór  — d dl 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mZ dd
lmZmZ ddl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) ddl*m+Z+ ddl,m-Z-m.Z.m/Z/ ddl0m1Z1m2Z2 ddl3m4Z4  G d„ dej5        ¦  «        Z6 ed¦  «         G d„ dej5        ¦  «        ¦   «         Z7d„ Z8 ed¦  «        dDd„¦   «         Z9dej:        d e;d!ej:        fd"„Z<	 dEd$ej5        d%ej:        d&ej:        d'ej:        d(ej:        dz  d)e=d*e=d+e&e-         fd,„Z> ee9¦  «         G d-„ d.ej5        ¦  «        ¦   «         Z?e G d/„ d0ej5        ¦  «        ¦   «         Z@ G d1„ d2ej5        ¦  «        ZA G d3„ d4ej5        ¦  «        ZB G d5„ d6ej5        ¦  «        ZC G d7„ d8e¦  «        ZDe( G d9„ d:e$¦  «        ¦   «         ZEe( G d;„ d<eE¦  «        ¦   «         ZF	 	 	 dFd>ej:        eGej:                 z  dz  d?e;dz  d(ej:        dz  d!ej:        e;z  fd@„ZHe( G dA„ dBeEe¦  «        ¦   «         ZIg dC¢ZJdS )Gé    )ÚCallableN)Únné   )Úinitialization)ÚACT2FN)ÚCacheÚDynamicCache)ÚGenerationMixin)Úuse_experts_implementationÚuse_kernel_forward_from_hubÚuse_kernelized_func)Úcreate_causal_maskÚ!create_sliding_window_causal_mask)ÚFlashAttentionKwargs)ÚGradientCheckpointingLayer)ÚMoeCausalLMOutputWithPastÚMoeModelOutputWithPast)ÚROPE_INIT_FUNCTIONSÚdynamic_rope_update)ÚALL_ATTENTION_FUNCTIONSÚPreTrainedModel)ÚUnpack)Úauto_docstringÚcan_return_tuple)Údeprecate_kwarg)ÚTransformersKwargsÚmaybe_autocastÚmerge_with_config_defaults)ÚOutputRecorderÚcapture_outputsé   )ÚMellumConfigc                   óú   ‡ — e Zd Z edd¬¦  «        ddefˆ fd„¦   «         Ze edd¬¦  «        	 ddededz  dee	j
        ef         fd	„¦   «         ¦   «         Z e	j        ¦   «         ed
„ ¦   «         ¦   «         Zˆ xZS )ÚMellumRotaryEmbeddingÚdevicez5.18)ÚversionNÚconfigc           	      ó¶  •— t          ¦   «                              ¦   «          |j        | _        |j        | _        || _        t          t          |j        ¦  «        ¦  «        | _        i | _	        | j        D ]ã}| j        j
        |         }|€Œ|d         | j	        |<   | j        }| j	        |         dk    rt          | j	        |                  } || j        ||¬¦  «        \  }}t          | |› d�t          j        |d¬¦  «        ¦  «         t          | |› d�t          j        |                     ¦   «         d¬¦  «        ¦  «         t          | |› d�|¦  «         Œäd S )	NÚ	rope_typeÚdefault©Ú
layer_typeÚ	_inv_freqF)Ú
persistentÚ_original_inv_freqÚ_attention_scaling)ÚsuperÚ__init__Úmax_position_embeddingsÚmax_seq_len_cachedÚoriginal_max_seq_lenr'   ÚsortedÚsetÚlayer_typesr)   Úrope_parametersÚcompute_default_rope_parametersr   Úsetattrr   ÚBufferÚclone)	Úselfr'   r%   r,   Úrope_paramsÚrope_init_fnÚcurr_inv_freqÚcurr_attention_scalingÚ	__class__s	           €úz/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/transformers/models/mellum/modeling_mellum.pyr2   zMellumRotaryEmbedding.__init__/   sl  ø€ å‰Œ×ÒÑÔÐØ"(Ô"@ˆÔØ$*Ô$BˆÔ!ØˆŒÝ!¥# fÔ&8Ñ"9Ô"9Ñ:Ô:ˆÔØˆŒØÔ*ð 	Uð 	UˆJØœ+Ô5°jÔAˆKØÐ"Øà)4°[Ô)AˆDŒN˜:Ñ&Ø%)Ô%IˆLØŒ~˜jÔ)¨YÒ6Ð6Ý2°4´>À*Ô3MÔN�Ø4@°LÀÄÈfÐakÐ4lÑ4lÔ4lÑ1ˆMÐ1Ý�D˜ZÐ2Ð2Ð2µB´I¸mÐX]Ð4^Ñ4^Ô4^Ñ_Ô_Ð_Ý�D˜ZÐ;Ð;Ð;½R¼YÀ}×GZÒGZÑG\ÔG\ÐinÐ=oÑ=oÔ=oÑpÔpÐpÝ�D˜ZÐ;Ð;Ð;Ð=SÑTÔTÐTÐTð	Uð 	Uó    r,   Úreturnc                 óV  — | j         |         d         }| j         |                              dd¦  «        }t          | dd¦  «        p| j        | j        z  }t          ||z  ¦  «        }d}d|t          j        d|d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.
            layer_type (`str`, *optional*):
                The current layer type if the model has different RoPE parameters per type.
                Should not be used unless `config.layer_types is not None`
        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partial_rotary_factorg      ð?Úhead_dimNr   é   ©Údtype)
r9   ÚgetÚgetattrÚhidden_sizeÚnum_attention_headsÚintÚtorchÚarangeÚfloatÚto)
r'   r%   r,   ÚkwargsÚbaserI   rJ   ÚdimÚattention_factorÚinv_freqs
             rD   r:   z5MellumRotaryEmbedding.compute_default_rope_parametersE   s´   € ð" Ô% jÔ1°,Ô?ˆà &Ô 6°zÔ B× FÒ FÐG^Ð`cÑ dÔ dÐÝ˜6 :¨tÑ4Ô4Ðh¸Ô8JÈfÔNhÑ8hˆÝ�(Ð2Ñ2Ñ3Ô3ˆàÐð ˜$¥5¤<°°3¸ÅÄÐ#MÑ#MÔ#MÐPSÑ#SÑTÑUˆØ�{Š{˜6Ñ"Ô"Ð$4Ð4Ð4rE   c                 ó(  — t          | |› d�¦  «        }t          | |› d�¦  «        }|d d d …d f                              |j        d         dd¦  «                             t          j        |j        ¬¦  «        }|d d …d d d …f                              ¦   «         }t          |j        j        t          ¦  «        r|j        j        dk    r|j        j        nd}t          |d	¬
¦  «        5  ||z                       dd¦  «        }	t	          j        |	|	fd¬¦  «        }
|
                     ¦   «         |z  }|
                     ¦   «         |z  }d d d ¦  «         n# 1 swxY w Y   |                     |j        ¬¦  «        |                     |j        ¬¦  «        fS )Nr-   r0   r   éÿÿÿÿr!   ©rM   r%   ÚmpsÚcpuF)Údevice_typeÚenabledrK   ©rY   rL   )rO   ÚexpandÚshaperV   rS   rU   r%   Ú
isinstanceÚtypeÚstrr   Ú	transposeÚcatÚcosÚsinrM   )r>   ÚxÚposition_idsr,   r[   Úattention_scalingÚinv_freq_expandedÚposition_ids_expandedra   ÚfreqsÚembrk   rl   s                rD   ÚforwardzMellumRotaryEmbedding.forwardb   sÈ  € õ ˜4 JÐ!9Ð!9Ð!9Ñ:Ô:ˆÝ# D¨ZÐ*KÐ*KÐ*KÑLÔLÐð �T˜1˜1˜1˜d�]Ô#×*Ò*¨<Ô+=¸aÔ+@À"ÀaÑHÔH×KÒKÕRWÔR]ÐfgÔfnÐKÑoÔoð 	ð !-¨Q¨Q¨Q°°a°a°a¨ZÔ 8× >Ò >Ñ @Ô @Ðå'1°!´(´-ÅÑ'EÔ'EÐkÈ!Ì(Ì-Ð[`ÒJ`ÐJ`�a”h”m�mÐfkˆå¨¸UÐCÑCÔCð 	0ð 	0Ø&Ð)>Ñ>×IÒIÈ!ÈQÑOÔOˆEÝ”)˜U E˜N°Ð3Ñ3Ô3ˆCØ—'’'‘)”)Ð/Ñ/ˆCØ—'’'‘)”)Ð/Ñ/ˆCð		0ð 	0ð 	0ñ 	0ô 	0ð 	0ð 	0ð 	0ð 	0ð 	0ð 	0øøøð 	0ð 	0ð 	0ð 	0ð �vŠv˜AœGˆvÑ$Ô$ c§f¢f°1´7 fÑ&;Ô&;Ð;Ð;s   Ã'A EÅEÅE©N)NN)Ú__name__Ú
__module__Ú__qualname__r   r"   r2   Ústaticmethodrh   ÚtuplerS   ÚTensorrU   r:   Úno_gradr   rt   Ú__classcell__©rC   s   @rD   r$   r$   .   s  ø€ € € € € Ø€_�X vÐ.Ñ.Ô.ðUð U˜|ð Uð Uð Uð Uð Uñ /Ô.ðUð* Ø€_�X vÐ.Ñ.Ô.àDHð5ð 5Øð5Ø7:¸T±zð5à	ˆuŒ|˜UÐ"Ô	#ð5ð 5ð 5ñ /Ô.ñ „\ð5ð6 €U„]�_„_Øð<ð <ñ Ôñ „_ð<ð <ð <ð <ð <rE   r$   ÚRMSNormc                   óT   ‡ — e Zd Zd	deddfˆ fd„Zdej        dej        fd„Zd„ Zˆ xZ	S )
ÚMellumRMSNormç�íµ ÷Æ°>ÚepsrF   Nc                 ó¬   •— t          ¦   «                              ¦   «          t          j        t	          j        |¦  «        ¦  «        | _        || _        dS )z<
        MellumRMSNorm is equivalent to T5LayerNorm
        N)r1   r2   r   Ú	ParameterrS   ÚonesÚweightÚvariance_epsilon)r>   rP   rƒ   rC   s      €rD   r2   zMellumRMSNorm.__init__z   sD   ø€ õ 	‰Œ×ÒÑÔÐÝ”l¥5¤:¨kÑ#:Ô#:Ñ;Ô;ˆŒØ #ˆÔÐÐrE   Úhidden_statesc                 ó  — |j         }|                     t          j        ¦  «        }|                     d¦  «                             dd¬¦  «        }|t          j        || j        z   ¦  «        z  }| j        |                     |¦  «        z  S )NrK   r]   T)Úkeepdim)	rM   rV   rS   Úfloat32ÚpowÚmeanÚrsqrtrˆ   r‡   )r>   r‰   Úinput_dtypeÚvariances       rD   rt   zMellumRMSNorm.forward‚   s|   € Ø#Ô)ˆØ%×(Ò(­¬Ñ7Ô7ˆØ ×$Ò$ QÑ'Ô'×,Ò,¨R¸Ð,Ñ>Ô>ˆØ%­¬°H¸tÔ?TÑ4TÑ(UÔ(UÑUˆØŒ{˜]×-Ò-¨kÑ:Ô:Ñ:Ð:rE   c                 óH   — t          | j        j        ¦  «        › d| j        › �S )Nz, eps=)rz   r‡   re   rˆ   )r>   s    rD   Ú
extra_reprzMellumRMSNorm.extra_repr‰   s&   € Ý˜œÔ)Ñ*Ô*ÐIÐI°$Ô2GÐIÐIÐIrE   )r‚   )
rv   rw   rx   rU   r2   rS   r{   rt   r“   r}   r~   s   @rD   r�   r�   x   sŒ   ø€ € € € € ð$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rE   r�   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]   rK   rc   )re   rS   rj   )rm   Úx1Úx2s      rD   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'rE   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krk   rl   Úunsqueeze_dimÚq_embedÚk_embeds          rD   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rE   r‰   Ún_reprF   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)re   rd   Úreshape)r‰   r¡   ÚbatchÚnum_key_value_headsÚslenrJ   s         rD   Ú	repeat_kvr§   ®   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TrE   ç        ÚmoduleÚqueryÚkeyÚvalueÚattention_maskÚscalingÚdropoutrW   c                 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 )NrK   r   r]   )rY   rM   )ÚpÚtrainingr!   )r§   Únum_key_value_groupsrS   Úmatmulri   r   Ú
functionalÚsoftmaxrŒ   rV   rM   r¯   r²   Ú
contiguous)r©   rª   r«   r¬   r­   r®   r¯   rW   Ú
key_statesÚvalue_statesÚattn_weightsÚattn_outputs               rD   Úeager_attention_forwardr¼   º   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$Ð$rE   c                   óÊ   ‡ — e Zd ZdZdede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         de	ej        ej        dz  f         fd„Zˆ xZS )ÚMellumAttentionz=Multi-headed attention from 'Attention Is All You Need' paperr'   Ú	layer_idxc                 ól  •— 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        ¬¦  «        | _        t)          | j        |j        ¬¦  «        | _        t)          | j        |j        ¬¦  «        | _        |j        |         dk    r|j        nd | _        d S )NrJ   g      à¿T©Úbias©rƒ   Úsliding_attention)r1   r2   r'   r¿   rO   rP   rQ   rJ   r¥   r³   r®   Úattention_dropoutÚ	is_causalr   ÚLinearÚattention_biasÚq_projÚk_projÚv_projÚo_projr�   Úrms_norm_epsÚq_normÚk_normr8   Úsliding_window©r>   r'   r¿   rC   s      €rD   r2   zMellumAttention.__init__×   sœ  ø€ Ý‰Œ×ÒÑÔÐØˆŒØ"ˆŒÝ ¨
°FÔ4FÈ&ÔJdÑ4dÑeÔeˆŒØ$*Ô$>À&ÔB\Ñ$\ˆÔ!Ø”} dÑ*ˆŒØ!'Ô!9ˆÔØˆŒå”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ Ô :¸T¼]Ñ JÐQWÔQfð
ñ 
ô 
ˆŒõ ”iØÔ&¨¬Ñ6¸Ô8JÐQWÔQfð
ñ 
ô 
ˆŒõ $ D¤M°vÔ7JÐKÑKÔKˆŒÝ# D¤M°vÔ7JÐKÑKÔKˆŒØ7=Ô7IÈ)Ô7TÐXkÒ7kÐ7k˜fÔ3Ð3ÐquˆÔÐÐrE   Nr‰   Úposition_embeddingsr­   Úpast_key_valuesrW   rF   c                 óz  — |j         d d…         }g |¢d‘| j        ‘R }|                      |                      |¦  «                             |¦  «        ¦  «                             dd¦  «        }|                      |                      |¦  «                             |¦  «        ¦  «                             dd¦  «        }	|                      |¦  «                             |¦  «                             dd¦  «        }
|\  }}t          ||	||¦  «        \  }}	|�| 
                    |	|
| j        ¦  «        \  }	}
t          j        | j        j        t           ¦  «        } || ||	|
|f| j        sdn| j        | j        | j        dœ|¤Ž\  }} |j        g |¢d‘R Ž                      ¦   «         }|                      |¦  «        }||fS )Nr]   r!   rK   r¨   )r¯   r®   rÐ   )re   rJ   rÎ   rÉ   Úviewri   rÏ   rÊ   rË   r    Úupdater¿   r   Úget_interfacer'   Ú_attn_implementationr¼   r²   rÅ   r®   rÐ   r£   r·   rÌ   )r>   r‰   rÒ   r­   rÓ   rW   Úinput_shapeÚhidden_shapeÚquery_statesr¸   r¹   rk   rl   Úattention_interfacer»   rº   s                   rD   rt   zMellumAttention.forwardñ   sà  € 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˜å(?Ô(MØŒKÔ,Õ.Eñ)
ô )
Ðð %8Ð$7ØØØØØð
%
ð  $œ}ÐH�C�C°$Ô2HØ”LØÔ.ð
%
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(rE   ru   )rv   rw   rx   Ú__doc__r"   rR   r2   rS   r{   rz   r   r   r   rt   r}   r~   s   @rD   r¾   r¾   Ó   så   ø€ € € € € àGÐGðv˜|ð v¸ð vð vð vð vð vð vð> )-ð')ð ')à”|ð')ð # 5¤<°´Ð#=Ô>ð')ð œ tÑ+ð	')ð
  ™ð')ð Ð-Ô.ð')ð 
ˆuŒ|˜Uœ\¨DÑ0Ð0Ô	1ð')ð ')ð ')ð ')ð ')ð ')ð ')ð ')rE   r¾   c                   ób   ‡ — e Zd ZdZˆ fd„Zdej        dej        dej        dej        fd„Zˆ xZS )ÚMellumExpertsz2Collection of expert weights stored as 3D tensors.c                 ó´  •— t          ¦   «                              ¦   «          |j        | _        |j        | _        |j        | _        t          j        t          j
        | j        d| j        z  | j        ¦  «        ¦  «        | _        t          j        t          j
        | j        | j        | j        ¦  «        ¦  «        | _        t          |j                 | _        d S )NrK   )r1   r2   Únum_expertsrP   Ú
hidden_dimÚmoe_intermediate_sizeÚintermediate_dimr   r…   rS   ÚemptyÚgate_up_projÚ	down_projr   Ú
hidden_actÚact_fn©r>   r'   rC   s     €rD   r2   zMellumExperts.__init__  s£   ø€ Ý‰Œ×ÒÑÔÐØ!Ô-ˆÔØ Ô,ˆŒØ &Ô <ˆÔÝœL­¬°TÔ5EÀqÈ4ÔK`ÑG`ÐbfÔbqÑ)rÔ)rÑsÔsˆÔÝœ¥e¤k°$Ô2BÀDÄOÐUYÔUjÑ&kÔ&kÑlÔlˆŒÝ˜VÔ.Ô/ˆŒˆˆrE   r‰   Útop_k_indexÚtop_k_weightsrF   c                 ó€  — t          j        |¦  «        }t          j        ¦   «         5  t           j        j                             || j        ¬¦  «        }|                     ddd¦  «        }t          j        | 	                    d¬¦  «        d¦  «         
                    ¦   «         }d d d ¦  «         n# 1 swxY w Y   |D ]þ}|d         }|| j        k    rŒt          j        ||         ¦  «        \  }}	||	         }
t          j                             |
| j        |         ¦  «                             dd¬¦  «        \  }}|                      |¦  «        |z  }t          j                             || j        |         ¦  «        }|||	|d f         z  }|                     d|	|                     |j        ¦  «        ¦  «         Œÿ|S )N)Únum_classesrK   r!   r   )r]   éþÿÿÿrc   r]   )rS   Ú
zeros_liker|   r   rµ   Úone_hotrá   ÚpermuteÚgreaterÚsumÚnonzeroÚwhereÚlinearræ   Úchunkré   rç   Ú
index_add_rV   rM   )r>   r‰   rë   rì   Úfinal_hidden_statesÚexpert_maskÚ
expert_hitÚ
expert_idxÚ	top_k_posÚ	token_idxÚcurrent_stateÚgateÚupÚcurrent_hidden_statess                 rD   rt   zMellumExperts.forward(  sø  € õ $Ô.¨}Ñ=Ô=ÐÝŒ]‰_Œ_ð 	Sð 	SÝœ(Ô-×5Ò5°kÈtÔO_Ð5Ñ`Ô`ˆKØ%×-Ò-¨a°°AÑ6Ô6ˆKÝœ {§¢¸8 Ñ'DÔ'DÀaÑHÔH×PÒPÑRÔRˆJð	Sð 	Sð 	Sñ 	Sô 	Sð 	Sð 	Sð 	Sð 	Sð 	Sð 	Søøøð 	Sð 	Sð 	Sð 	Sð
 %ð 
	nð 
	nˆJØ# AœˆJØ˜TÔ-Ò-Ð-ØÝ#(¤;¨{¸:Ô/FÑ#GÔ#GÑ ˆI�yØ)¨)Ô4ˆMÝ”}×+Ò+¨M¸4Ô;LÈZÔ;XÑYÔY×_Ò_Ð`aÐgiÐ_ÑjÔj‰HˆD�"Ø$(§K¢K°Ñ$5Ô$5¸Ñ$:Ð!Ý$&¤M×$8Ò$8Ð9NÐPTÔP^Ð_iÔPjÑ$kÔ$kÐ!Ø$9¸MÈ)ÐU^Ð`dÐJdÔ<eÑ$eÐ!Ø×*Ò*¨1¨iÐ9N×9QÒ9QÐReÔRkÑ9lÔ9lÑmÔmÐmÐmà"Ð"s   ¨A>B2Â2B6Â9B6)	rv   rw   rx   rÝ   r2   rS   r{   rt   r}   r~   s   @rD   rß   rß     s€   ø€ € € € € à<Ð<ð0ð 0ð 0ð 0ð 0ð#à”|ð#ð ”\ð#ð ”|ð	#ð
 
Œð#ð #ð #ð #ð #ð #ð #ð #rE   rß   c                   ó$   ‡ — e Zd Zˆ fd„Zd„ Zˆ xZS )ÚMellumTopKRouterc                 ó  •— t          ¦   «                              ¦   «          |j        | _        |j        | _        |j        | _        |j        | _        t          j	        t          j        | j        | j        ¦  «        ¦  «        | _        d S ru   )r1   r2   Únum_experts_per_tokÚtop_krá   Únorm_topk_probrP   râ   r   r…   rS   Úzerosr‡   rê   s     €rD   r2   zMellumTopKRouter.__init__D  si   ø€ Ý‰Œ×ÒÑÔÐØÔ/ˆŒ
Ø!Ô-ˆÔØ$Ô3ˆÔØ Ô,ˆŒÝ”l¥5¤;¨tÔ/?ÀÄÑ#QÔ#QÑRÔRˆŒˆˆrE   c                 ó�  — |                      d| j        ¦  «        }t          j        || j        ¦  «        }t
          j        j                             |t
          j	        d¬¦  «        }t          j
        || j        d¬¦  «        \  }}| j        r||                     dd¬¦  «        z  }|                     |j        ¦  «        }|}|||fS )Nr]   )rM   rY   rc   T)rY   r‹   )r£   râ   ÚFr÷   r‡   rS   r   rµ   r¶   rU   Útopkr  r	  rô   rV   rM   )r>   r‰   Úrouter_logitsÚrouter_probsÚrouter_top_valueÚrouter_indicesÚrouter_scoress          rD   rt   zMellumTopKRouter.forwardL  sÁ   € Ø%×-Ò-¨b°$´/ÑBÔBˆÝœ °´Ñ<Ô<ˆÝ”xÔ*×2Ò2°=ÍÌÐY[Ð2Ñ\Ô\ˆÝ+0¬:°lÀDÄJÐTVÐ+WÑ+WÔ+WÑ(Ð˜.ØÔð 	KØÐ 0× 4Ò 4¸ÀTÐ 4Ñ JÔ JÑJÐØ+×.Ò.¨}Ô/BÑCÔCÐØ(ˆØ˜m¨^Ð;Ð;rE   ©rv   rw   rx   r2   rt   r}   r~   s   @rD   r  r  C  sL   ø€ € € € € ðSð Sð Sð Sð Sð	<ð 	<ð 	<ð 	<ð 	<ð 	<ð 	<rE   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MellumSparseMoeBlockr'   c                 ó˜   •— t          ¦   «                              ¦   «          t          |¦  «        | _        t	          |¦  «        | _        d S ru   )r1   r2   rß   Úexpertsr  r  rê   s     €rD   r2   zMellumSparseMoeBlock.__init__Y  s;   ø€ Ý‰Œ×ÒÑÔÐÝ$ VÑ,Ô,ˆŒÝ$ VÑ,Ô,ˆŒ	ˆ	ˆ	rE   r‰   rF   c                 óÒ   — |j         \  }}}|                     d|¦  «        }|                      |¦  «        \  }}}|                      |||¦  «        }	|	                     |||¦  «        S )Nr]   )re   rÕ   r  r  r£   )
r>   r‰   Ú
batch_sizeÚsequence_lengthrâ   Úhidden_states_reshapedÚ_Úrouting_weightsÚselected_expertsrú   s
             rD   rt   zMellumSparseMoeBlock.forward^  sr   € Ø2?Ô2EÑ/ˆ
�O ZØ!.×!3Ò!3°B¸
Ñ!CÔ!CÐØ/3¯yªyÐ9OÑ/PÔ/PÑ,ˆˆ?Ð,Ø"ŸlšlÐ+AÐCSÐUdÑeÔeÐØ"×*Ò*¨:°È
ÑSÔSÐSrE   )	rv   rw   rx   r"   r2   rS   r{   rt   r}   r~   s   @rD   r  r  X  st   ø€ € € € € ð-˜|ð -ð -ð -ð -ð -ð -ð
T U¤\ð T°e´lð Tð Tð Tð Tð Tð Tð Tð TrE   r  c                   ó&   ‡ — e Zd Zdˆ fd„	Zd„ Zˆ xZS )Ú	MellumMLPNc                 ó   •— t          ¦   «                              ¦   «          || _        |j        | _        |€|j        n|| _        t          j        | j        | j        d¬¦  «        | _        t          j        | j        | j        d¬¦  «        | _        t          j        | j        | j        d¬¦  «        | _	        t          |j                 | _        d S ©NFrÁ   )r1   r2   r'   rP   Úintermediate_sizer   rÇ   Ú	gate_projÚup_projrç   r   rè   ré   )r>   r'   r#  rC   s      €rD   r2   zMellumMLP.__init__g  s±   ø€ Ý‰Œ×ÒÑÔÐØˆŒØ!Ô-ˆÔØ=NÐ=V Ô!9Ð!9Ð\mˆÔÝœ 4Ô#3°TÔ5KÐRWÐXÑXÔXˆŒÝ”y Ô!1°4Ô3IÐPUÐVÑVÔVˆŒÝœ 4Ô#9¸4Ô;KÐRWÐXÑXÔXˆŒÝ˜VÔ.Ô/ˆŒˆˆrE   c                 ó¨   — |                       |                      |                      |¦  «        ¦  «        |                      |¦  «        z  ¦  «        }|S ru   )rç   ré   r$  r%  )r>   rm   rç   s      rD   rt   zMellumMLP.forwardq  sA   € Ø—N’N 4§;¢;¨t¯~ª~¸aÑ/@Ô/@Ñ#AÔ#AÀDÇLÂLÐQRÁOÄOÑ#SÑTÔTˆ	ØÐrE   ru   r  r~   s   @rD   r   r   f  sL   ø€ € € € € ð0ð 0ð 0ð 0ð 0ð 0ðð ð ð ð ð ð rE   r   c                   óÒ   ‡ — 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
dz  deej        ej        f         dz  dee         dej        fd„Zˆ xZS )ÚMellumDecoderLayerr'   r¿   c                 óŒ  •— t          ¦   «                              ¦   «          |j        | _        t          ||¦  «        | _        |j        |         dk    rt          |¦  «        | _        nt          ||j	        ¬¦  «        | _        t          |j        |j        ¬¦  «        | _        t          |j        |j        ¬¦  «        | _        d S )NÚsparse)r#  rÃ   )r1   r2   rP   r¾   Ú	self_attnÚmlp_layer_typesr  Úmlpr   r#  r�   rÍ   Úinput_layernormÚpost_attention_layernormrÑ   s      €rD   r2   zMellumDecoderLayer.__init__w  s¬   ø€ Ý‰Œ×ÒÑÔÐØ!Ô-ˆÔÝ(¨°Ñ;Ô;ˆŒØÔ! )Ô,°Ò8Ð8Ý+¨FÑ3Ô3ˆDŒHˆHå  ¸6Ô;SÐTÑTÔTˆDŒHÝ,¨VÔ-?ÀVÔEXÐYÑYÔYˆÔÝ(5°fÔ6HÈfÔNaÐ(bÑ(bÔ(bˆÔ%Ð%Ð%rE   NFr‰   r­   rn   rÓ   Ú	use_cacherÒ   rW   rF   c           
      óÎ   — |}|                       |¦  «        } | j        d||||||dœ|¤Ž\  }}	||z   }|}|                      |¦  «        }|                      |¦  «        }||z   }|S )N)r‰   r­   rn   rÓ   r0  rÒ   © )r.  r+  r/  r-  )
r>   r‰   r­   rn   rÓ   r0  rÒ   rW   Úresidualr  s
             rD   rt   zMellumDecoderLayer.forward‚  s¡   € ð !ˆØ×,Ò,¨]Ñ;Ô;ˆà)˜4œ>ð 
Ø'Ø)Ø%Ø+ØØ 3ð
ð 
ð ð
ð 
Ñˆ�qð ! =Ñ0ˆð !ˆØ×5Ò5°mÑDÔDˆØŸš Ñ/Ô/ˆØ  =Ñ0ˆØÐrE   )NNNFN)rv   rw   rx   r"   rR   r2   rS   r{   Ú
LongTensorr   Úboolrz   r   r   rt   r}   r~   s   @rD   r(  r(  v  sÿ   ø€ € € € € ð	c˜|ð 	c¸ð 	cð 	cð 	cð 	cð 	cð 	cð /3Ø04Ø(,Ø!&ØHLðð à”|ðð œ tÑ+ðð Ô&¨Ñ-ð	ð
  ™ðð ˜$‘;ðð # 5¤<°´Ð#=Ô>ÀÑEðð Ð+Ô,ðð 
Œðð ð ð ð ð ð ð rE   r(  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MellumPreTrainedModelr'   ÚmodelTr(  rÓ   r   )Úindex)r  r‰   Ú
attentionsc                 óÄ  •— t          ¦   «                              |¦  «         | j        j        }t	          |t
          ¦  «        r9t          j        |j        d|¬¦  «         t          j        |j	        d|¬¦  «         n1t	          |t          ¦  «        rt          j        |j        d|¬¦  «         t	          |t          ¦  «        r›|j        D ]•}|j        }|j        |         dk    rt           |j        |                  } ||j        |¬¦  «        \  }}t          j        t%          ||› d�¦  «        |¦  «         t          j        t%          ||› d�¦  «        |¦  «         Œ”d S d S )Nr¨   )rŽ   Ústdr*   r+   r-   r/   )r1   Ú_init_weightsr'   Úinitializer_rangerf   rß   ÚinitÚnormal_ræ   rç   r  r‡   r$   r8   r:   r)   r   Úcopy_rO   )r>   r©   r<  r,   r@   rA   r  rC   s          €rD   r=  z#MellumPreTrainedModel._init_weightsµ  so  ø€ å‰Œ×Ò˜fÑ%Ô%Ð%ØŒkÔ+ˆÝ�f�mÑ,Ô,ð 	;ÝŒL˜Ô,°3¸CÐ@Ñ@Ô@Ð@ÝŒL˜Ô)°¸Ð=Ñ=Ô=Ð=Ð=Ý˜Õ 0Ñ1Ô1ð 	;ÝŒL˜œ¨S°cÐ:Ñ:Ô:Ð:Ý�fÕ3Ñ4Ô4ð 	^Ø$Ô0ð ^ð ^�
Ø%ÔE�ØÔ# JÔ/°9Ò<Ð<Ý#6°vÔ7GÈ
Ô7SÔ#T�LØ#/ <°´È*Ð#UÑ#UÔ#UÑ �˜qÝ”
�7 6¨jÐ+CÐ+CÐ+CÑDÔDÀmÑTÔTÐTÝ”
�7 6¨jÐ+LÐ+LÐ+LÑMÔMÈ}Ñ]Ô]Ð]Ð]ð	^ð 	^ð^ð ^rE   )rv   rw   rx   r"   Ú__annotations__Úbase_model_prefixÚsupports_gradient_checkpointingÚ_no_split_modulesÚ_skip_keys_device_placementÚ_supports_flash_attnÚ_supports_sdpaÚ_supports_flex_attnÚ_can_compile_fullgraphÚ_supports_attention_backendr   r  r(  r¾   Ú_can_record_outputsrS   r|   r=  r}   r~   s   @rD   r7  r7  ¢  sÄ   ø€ € € € € € àÐÐÑØÐØ&*Ð#Ø-Ð.ÐØ#4Ð"5ÐØÐØ€NØÐà!ÐØ"&Ðà'˜Ð(8ÀÐBÑBÔBØ+Ø%ðð Ðð €U„]�_„_ð^ð ^ð ^ð ^ñ „_ð^ð ^ð ^ð ^ð ^rE   r7  c                   óâ   ‡ — e Zd Zdefˆ fd„Zeee	 	 	 	 	 	 ddej	        dz  dej
        dz  dej	        dz  dedz  dej        dz  d	edz  d
ee         defd„¦   «         ¦   «         ¦   «         Zˆ xZS )ÚMellumModelr'   c                 óÞ  •‡— t          ¦   «                              ‰¦  «         ‰j        | _        ‰j        | _        t          j        ‰j        ‰j        | j        ¦  «        | _        t          j	        ˆfd„t          ‰j        ¦  «        D ¦   «         ¦  «        | _        t          ‰j        ‰j        ¬¦  «        | _        t!          ‰¬¦  «        | _        d| _        |                      ¦   «          d S )Nc                 ó0   •— g | ]}t          ‰|¦  «        ‘ŒS r2  )r(  )Ú.0r¿   r'   s     €rD   ú
<listcomp>z(MellumModel.__init__.<locals>.<listcomp>Ñ  s$   ø€ ÐdÐdÐd°yÕ ¨	Ñ2Ô2ÐdÐdÐdrE   rÃ   ©r'   F)r1   r2   Úpad_token_idÚpadding_idxÚ
vocab_sizer   Ú	EmbeddingrP   Úembed_tokensÚ
ModuleListÚrangeÚnum_hidden_layersÚlayersr�   rÍ   Únormr$   Ú
rotary_embÚgradient_checkpointingÚ	post_initrê   s    `€rD   r2   zMellumModel.__init__Ê  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dÐdÐdÐdÅEÈ&ÔJbÑDcÔDcÐdÑdÔdñ
ô 
ˆŒõ " &Ô"4¸&Ô:MÐNÑNÔNˆŒ	Ý/°vÐ>Ñ>Ô>ˆŒØ&+ˆÔ#ð 	�ŠÑÔÐÐÐrE   NÚ	input_idsr­   rn   rÓ   Úinputs_embedsr0  rW   rF   c           	      ó˜  ‡— |d u |d uz  rt          d¦  «        ‚|€|                      |¦  «        }|r|€t          | j        ¬¦  «        }|€V|�|                     ¦   «         nd}t          j        |j        d         |j        ¬¦  «        |z   }| 	                    d¦  «        }t          |x}	t          ¦  «        sI| j        ||||dœŠˆfd„ˆfd„d	œ}
i }	t          | j        j        ¦  «        D ]} |
|         ¦   «         |	|<   Œ|}i }t          | j        j        ¦  «        D ]}|                      |||¦  «        ||<   Œt          | j        d | j        j        …         ¦  «        D ]?\  }} ||f|	| j        j        |                  || j        j        |                  ||d
œ|¤Ž}Œ@|                      |¦  «        }t'          ||r|nd ¬¦  «        S )Nz:You must specify exactly one of input_ids or inputs_embedsrS  r   r!   )r%   )r'   rb  r­   rÓ   rn   c                  ó   •— t          di ‰ ¤ŽS ©Nr2  )r   ©Úmask_kwargss   €rD   ú<lambda>z%MellumModel.forward.<locals>.<lambda>þ  s   ø€ Õ*<Ð*KÐ*K¸{Ð*KÐ*K€ rE   c                  ó   •— t          di ‰ ¤ŽS re  )r   rf  s   €rD   rh  z%MellumModel.forward.<locals>.<lambda>ÿ  s   ø€ Õ-NÐ-]Ð-]ÐQ\Ð-]Ð-]€ rE   )Úfull_attentionrÄ   )r­   rÒ   rn   rÓ   )Úlast_hidden_staterÓ   )Ú
ValueErrorrX  r	   r'   Úget_seq_lengthrS   rT   re   r%   rš   rf   Údictr7   r8   r^  Ú	enumerater\  r[  r]  r   )r>   ra  r­   rn   rÓ   rb  r0  rW   Úpast_seen_tokensÚcausal_mask_mappingÚmask_creation_functionsr,   r‰   rÒ   ÚiÚdecoder_layerrg  s                   @rD   rt   zMellumModel.forwardÚ  sc  ø€ ð ˜Ð -°tÐ";Ñ<ð 	[ÝÐYÑZÔZÐZàÐ Ø ×-Ò-¨iÑ8Ô8ˆMàð 	?˜Ð0Ý*°$´+Ð>Ñ>Ô>ˆOàÐØCRÐC^˜×=Ò=Ñ?Ô?Ð?ÐdeÐÝ œ<¨Ô(;¸AÔ(>À}ÔG[Ð\Ñ\Ô\Ð_oÑoˆLØ'×1Ò1°!Ñ4Ô4ˆLå°Ð?Ð-ÅÑFÔFð 	Xàœ+Ø!.Ø"0Ø#2Ø ,ðð ˆKð #LÐ"KÐ"KÐ"KØ%]Ð%]Ð%]Ð%]ð'ð 'Ð#ð #%ÐÝ! $¤+Ô"9Ñ:Ô:ð Xð X�
Ø2UÐ2IÈ*Ô2UÑ2WÔ2WÐ# JÑ/Ð/à%ˆØ ÐÝ˜dœkÔ5Ñ6Ô6ð 	gð 	gˆJØ.2¯oªo¸mÈ\Ð[eÑ.fÔ.fÐ 
Ñ+Ð+å )¨$¬+Ð6U¸¼Ô8UÐ6UÔ*VÑ WÔ Wð 	ð 	ÑˆAˆ}Ø)˜MØðà2°4´;Ô3JÈ1Ô3MÔNØ$7¸¼Ô8OÐPQÔ8RÔ$SØ)Ø /ðð ð ðð ˆMˆMð Ÿ	š	 -Ñ0Ô0ˆå%Ø+Ø/8ÐB˜O˜O¸dð
ñ 
ô 
ð 	
rE   )NNNNNN)rv   rw   rx   r"   r2   r   r    r   rS   r4  r{   r   ÚFloatTensorr5  r   r   r   rt   r}   r~   s   @rD   rN  rN  È  s  ø€ € € € € ð˜|ð ð ð ð ð ð ð   ØØð .2Ø.2Ø04Ø(,Ø26Ø!%ð<
ð <
àÔ# dÑ*ð<
ð œ tÑ+ð<
ð Ô&¨Ñ-ð	<
ð
  ™ð<
ð Ô(¨4Ñ/ð<
ð ˜$‘;ð<
ð Ð+Ô,ð<
ð 
 ð<
ð <
ð <
ñ „^ñ „_ñ  Ôð<
ð <
ð <
ð <
ð <
rE   rN  rK   Úgate_logitsrá   c           	      óÎ  — | �t          | t          ¦  «        sdS | d         j        }t          j        |t          j        |¬¦  «        }t          j        |t          j        |¬¦  «        }d}|�4|                     d¦  «                             |t          j        ¬¦  «        }| D �]‰}	t          j        j	         
                    |	                     |¦  «        d¬¦  «        }
t          j        |
|d¬¦  «        \  }}|€z|t          j        |                     d¦  «        |¬¦  «                             ¦   «         z   }||
                     ¦   «                              d¬¦  «        z   }||
j        d         z   }ŒÒ|t          j        |t          j        |¬¦  «                             d|                     d¦  «        |                     |¦  «        ¦  «        z   }||
                     ¦   «         |                     d¦  «        z                       d¬¦  «        z   }||                     ¦   «         z   }�Œ‹||z  }||z  }t          j        ||                     d¦  «        z  ¦  «        }||z  S )	aÄ  
    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 `gate`, should be a tuple of model.config.num_hidden_layers tensors of
            shape [batch_size X sequence_length, num_experts].
        num_experts:
            Number of experts
        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^   r¨   r]   )r%   rM   rc   )Ú	minlength)rf   rz   r%   rS   r
  rŒ   r£   rV   r   rµ   r¶   r  ÚbincountrU   rô   re   Úscatter_add_Úrepeat_interleaverš   )rv  rá   r  r­   Úcompute_deviceÚtokens_per_expert_sumÚrouter_prob_sumÚ
total_rowsÚ	flat_maskÚ
layer_gater  r  r  Útokens_per_expertÚrouter_prob_per_expertÚoverall_losss                   rD   Úload_balancing_loss_funcr…    sV  € ð: Ð¥*¨[½%Ñ"@Ô"@ÐØˆqð ! ”^Ô*€NÝ!œK¨½5¼=ÐQ_Ð`Ñ`Ô`ÐÝ”k +µU´]È>ÐZÑZÔZ€OØ€JàÐ!à"×*Ò*¨2Ñ.Ô.×1Ò1¸ÍuÌ}Ð1Ñ]Ô]ˆ	à!ð 6ñ 6ˆ
Ýœ(Ô-×5Ò5°j·m²mÀNÑ6SÔ6SÐY[Ð5Ñ\Ô\ˆÝ#œj¨¸%ÀRÐHÑHÔHÑˆÐØÐ!ð &­¬Ð7G×7OÒ7OÐPRÑ7SÔ7SÐ_jÐ(kÑ(kÔ(k×(qÒ(qÑ(sÔ(sÑsð "ð .°×0EÒ0EÑ0GÔ0G×0KÒ0KÐPQÐ0KÑ0RÔ0RÑRˆOØ# oÔ&;¸AÔ&>Ñ>ˆJˆJð %:½E¼KØ¥5¤=¸ð=ñ =ô =çŠl˜1Ð.×6Ò6°rÑ:Ô:¸I×<WÒ<WÐX]Ñ<^Ô<^Ñ_Ô_ñ%`Ð!ð .°×1FÒ1FÑ1HÔ1HÈ9×K^ÒK^Ð_aÑKbÔKbÑ1b×0gÒ0gÐlmÐ0gÑ0nÔ0nÑnˆOØ# i§m¢m¡o¤oÑ5ˆJ‰Jà-°
Ñ:ÐØ,¨zÑ9Ðå”9Ð.Ð1G×1QÒ1QÐRSÑ1TÔ1TÑTÑUÔU€LØ˜+Ñ%Ð%rE   c                   ó,  ‡ — e Zd ZddiZddiZddgdgfiZdd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d	z  dee
j        z  dee         defd„¦   «         ¦   «         Zˆ xZS )ÚMellumForCausalLMzlm_head.weightzmodel.embed_tokens.weightÚlm_headÚcolwise_gather_outputr‰   ÚlogitsÚkeep_full_weightc                 óF  •— t          ¦   «                              |¦  «         t          |¦  «        | _        |j        | _        t          j        |j        |j        d¬¦  «        | _        |j	        | _	        |j
        | _
        |j        | _        |                      ¦   «          d S r"  )r1   r2   rN  r8  rV  r   rÇ   rP   rˆ  Úrouter_aux_loss_coefrá   r  r`  rê   s     €rD   r2   zMellumForCausalLM.__init__h  s‹   ø€ Ý‰Œ×Ò˜Ñ Ô Ð Ý  Ñ(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rE   Nr   ra  r­   rn   rÓ   rb  Úlabelsr0  Úoutput_router_logitsÚlogits_to_keeprW   rF   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}|rHt          |j        | j        | j        |¦  «        }|�%|| j        |                     |j        ¦  «        z  z  }t#          ||||j        |j        |j        |j        ¬¦  «        S )ar  
        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, MellumForCausalLM

        >>> model = MellumForCausalLM.from_pretrained("Qwen/Qwen3-MoE-15B-A2B")
        >>> tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-MoE-15B-A2B")

        >>> 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)ra  r­   rn   rÓ   rb  r0  r�  )ÚlossÚaux_lossrŠ  rÓ   r‰   r:  r  r2  )r'   r�  r8  rk  rf   rR   Úslicerˆ  Úloss_functionrV  r…  r  rá   r  r�  rV   r%   r   rÓ   r‰   r:  )r>   ra  r­   rn   rÓ   rb  rŽ  r0  r�  r�  rW   Úoutputsr‰   Úslice_indicesrŠ  r’  r“  s                    rD   rt   zMellumForCausalLM.forwardt  sn  € ðN %9Ð$DÐ Ð È$Ì+ÔJjð 	ð
 +5¨$¬*ð 	+
ØØ)Ø%Ø+Ø'ØØ!5ð	+
ð 	+
ð ð	+
ð 	+
ˆð  Ô1ˆå8BÀ>ÕSVÑ8WÔ8WÐk�˜~˜o¨tÑ4Ô4Ð4Ð]kˆØ—’˜m¨A¨A¨A¨}¸a¸a¸aÐ,?Ô@ÑAÔAˆàˆØÐØ%�4Ô% f¨f°d´oÐPÐPÈÐPÐPˆDàˆØð 	MÝ/ØÔ%ØÔ ØÔ(Øñ	ô ˆHð Ð!Ø˜Ô1°H·K²KÀÄÑ4LÔ4LÑLÑL�å(ØØØØ#Ô3Ø!Ô/ØÔ)Ø!Ô/ð
ñ 
ô 
ð 	
rE   )	NNNNNNNNr   )rv   rw   rx   Ú_tied_weights_keysÚ_tp_planÚ_pp_planÚ
_fsdp_planr2   r   r   rS   r4  r{   r   ru  r5  rR   r   r   r   rt   r}   r~   s   @rD   r‡  r‡  a  s{  ø€ € € € € à*Ð,GÐHÐØÐ2Ð3€HØ˜_Ð-°¨zÐ:Ð;€HØÐ/Ð0€Jð
ð 
ð 
ð 
ð 
ð Øð .2Ø.2Ø04Ø(,Ø26Ø*.Ø!%Ø,0Ø-.ðP
ð P
àÔ# dÑ*ðP
ð œ tÑ+ðP
ð Ô&¨Ñ-ð	P
ð
  ™ðP
ð Ô(¨4Ñ/ðP
ð Ô  4Ñ'ðP
ð ˜$‘;ðP
ð # T™kðP
ð ˜eœlÑ*ðP
ð Ð+Ô,ðP
ð 
#ðP
ð P
ð P
ñ „^ñ ÔðP
ð P
ð P
ð P
ð P
rE   r‡  )r‡  rN  r7  )r!   )r¨   )NrK   N)KÚcollections.abcr   rS   Útorch.nn.functionalr   rµ   r  Ú r   r?  Úactivationsr   Úcache_utilsr   r	   Ú
generationr
   Úintegrationsr   r   r   Úmasking_utilsr   r   Úmodeling_flash_attention_utilsr   Úmodeling_layersr   Úmodeling_outputsr   r   Úmodeling_rope_utilsr   r   Úmodeling_utilsr   r   Úprocessing_utilsr   Úutilsr   r   Úutils.deprecationr   Úutils.genericr   r   r   Úutils.output_capturingr   r    Úconfiguration_mellumr"   ÚModuler$   r�   r—   r    r{   rR   r§   rU   r¼   r¾   rß   r  r  r   r(  r7  rN  rz   r…  r‡  Ú__all__r2  rE   rD   ú<module>r±     s½  ðð* %Ð $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 hÐ hÐ hÐ hÐ hÐ hÐ hÐ hÐ hÐ hØ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RØ BÐ BÐ BÐ BÐ BÐ BØ 9Ð 9Ð 9Ð 9Ð 9Ð 9Ø QÐ QÐ QÐ QÐ QÐ QÐ QÐ QØ KÐ KÐ KÐ KÐ KÐ KÐ KÐ KØ FÐ FÐ FÐ FÐ FÐ FÐ FÐ FØ &Ð &Ð &Ð &Ð &Ð &Ø 5Ð 5Ð 5Ð 5Ð 5Ð 5Ð 5Ð 5Ø 0Ð 0Ð 0Ð 0Ð 0Ð 0Ø [Ð [Ð [Ð [Ð [Ð [Ð [Ð [Ð [Ð [Ø EÐ EÐ EÐ EÐ EÐ EÐ EÐ EØ .Ð .Ð .Ð .Ð .Ð .ðG<ð G<ð G<ð G<ð G<˜BœIñ G<ô G<ð G<ðT Ð˜YÑ'Ô'ðJð Jð Jð Jð J�B”Iñ Jô Jñ (Ô'ðJð((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 %ð2 ÐÐ)Ñ*Ô*ðD)ð D)ð D)ð D)ð D)�b”iñ D)ô D)ñ +Ô*ðD)ðN ð$#ð $#ð $#ð $#ð $#�B”Iñ $#ô $#ñ Ôð$#ðN<ð <ð <ð <ð <�r”yñ <ô <ð <ð*Tð Tð Tð Tð T˜2œ9ñ Tô Tð Tðð ð ð ð �”	ñ ô ð ð )ð )ð )ð )ð )Ð3ñ )ô )ð )ðX ð"^ð "^ð "^ð "^ð "^˜Oñ "^ô "^ñ „ð"^ðJ ðP
ð P
ð P
ð P
ð P
Ð'ñ P
ô P
ñ „ðP
ðj #Ø
Ø*.ð	B&ð B&Ø”  e¤lÔ 3Ñ3°dÑ:ðB&à�t‘ðB&ð ”L 4Ñ'ð	B&ð
 „\�CÑðB&ð B&ð B&ð B&ðJ ðd
ð d
ð d
ð d
ð d
Ð-¨ñ d
ô d
ñ „ðd
ðN HÐ
GÐ
G€€€rE   