Ë
    ÜÓ^j^U  ã                   óð  — d dl mZ d dlmZ d dl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mZ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$m%Z% ddl&m'Z'm(Z(m)Z) ddl*m+Z+m,Z, ddl-m.Z. ddl/m0Z0 d„ Z1 ed«      dBd„«       Z2dejf                  de4dejf                  fd„Z5	 dCdejl                  dejf                  dejf                  d ejf                  d!ejf                  dz  d"e7d#e7d$e%e'   fd%„Z8d&ejf                  d'e7d(e4dejf                  fd)„Z9 ee2«       G d*„ d+ejl                  «      «       Z: G d,„ d-ejl                  «      Z; ed.«       G d/„ d0ejl                  «      «       Z< G d1„ d2e«      Z=e( G d3„ d4e#«      «       Z> G d5„ d6ejl                  «      Z?e( G d7„ d8e>«      «       Z@e( G d9„ d:e>e«      «       ZA G d;„ d<ee>«      ZB G d=„ d>ee>«      ZC G d?„ d@ee>«      ZDg dA¢ZEy)Dé    )ÚCallable)ÚOptionalN)Únné   )ÚACT2FN)ÚCacheÚDynamicCache)ÚGenerationMixin)Úuse_kernel_forward_from_hubÚuse_kernel_func_from_hubÚuse_kernelized_func)Úcreate_causal_maskÚ!create_sliding_window_causal_mask)ÚFlashAttentionKwargs)ÚGenericForQuestionAnsweringÚ GenericForSequenceClassificationÚGenericForTokenClassificationÚGradientCheckpointingLayer)ÚBaseModelOutputWithPastÚCausalLMOutputWithPast)ÚROPE_INIT_FUNCTIONSÚdynamic_rope_update)ÚALL_ATTENTION_FUNCTIONSÚPreTrainedModel)ÚUnpack)ÚTransformersKwargsÚauto_docstringÚcan_return_tuple)Úmaybe_autocastÚmerge_with_config_defaults)Úcapture_outputsé   )ÚMinistral3Configc                 óš   — | 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..Néÿÿÿÿé   ©Údim)ÚshapeÚtorchÚcat)ÚxÚx1Úx2s      ú}/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/transformers/models/ministral3/modeling_ministral3.pyÚrotate_halfr0   #   sZ   € à	
ˆ3Ð"�!—'‘'˜"‘+ Ñ"Ð"Ð"Ñ	#€BØ	
ˆ3�—‘˜‘˜qÑ Ñ"Ð"Ñ	#€BÜ�9‰9�r�c˜2�Y BÔ'Ð'ó    Úrotary_pos_embc                 óž   — |j                  |«      }|j                  |«      }| |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	unsqueezer0   )ÚqÚkÚcosÚsinÚunsqueeze_dimÚq_embedÚk_embeds          r/   Úapply_rotary_pos_embr<   *   sY   € ð& �-‰-˜Ó
&€CØ
�-‰-˜Ó
&€CØ�3‰wœ; q›>¨CÑ/Ñ0€GØ�3‰wœ; q›>¨CÑ/Ñ0€GØ�GÐÐr1   Úhidden_statesÚn_repÚreturnc                 óª   — | j                   \  }}}}|dk(  r| S | dd…dd…ddd…dd…f   j                  |||||«      } | j                  |||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)r)   ÚexpandÚreshape)r=   r>   ÚbatchÚnum_key_value_headsÚslenÚhead_dims         r/   Ú	repeat_kvrG   D   so   € ð
 2?×1DÑ1DÑ.€EÐ  hØ�‚zØÐØ!¢!¢Q¨ªa²Ð"2Ñ3×:Ñ:¸5ÐBUÐW\Ð^bÐdlÓm€MØ× Ñ  Ð(;¸eÑ(CÀTÈ8ÓTÐTr1   ÚmoduleÚqueryÚkeyÚvalueÚattention_maskÚscalingÚdropoutÚkwargsc                 ó  — t        || j                  «      }t        || j                  «      }	t        j                  ||j	                  dd«      «      |z  }
|�|
|z   }
t
        j                  j                  |
dt        j                  ¬«      j                  |j                  «      }
t
        j                  j                  |
|| j                  ¬«      }
t        j                  |
|	«      }|j	                  dd«      j                  «       }||
fS )Nr&   r   r%   )r(   Údtype)ÚpÚtrainingr"   )rG   Únum_key_value_groupsr*   ÚmatmulÚ	transposer   Ú
functionalÚsoftmaxÚfloat32ÚtorQ   rN   rS   Ú
contiguous)rH   rI   rJ   rK   rL   rM   rN   rO   Ú
key_statesÚvalue_statesÚattn_weightsÚattn_outputs               r/   Úeager_attention_forwardr`   P   sà   € ô ˜3 × ;Ñ ;Ó<€JÜ˜U F×$?Ñ$?Ó@€Lä—<‘<  z×';Ñ';¸A¸qÓ'AÓBÀWÑL€LØÐ!Ø# nÑ4ˆä—=‘=×(Ñ(¨¸2ÄUÇ]Á]Ð(ÓS×VÑVÐW\×WbÑWbÓc€LÜ—=‘=×(Ñ(¨¸È6Ï?É?Ð(Ó[€LÜ—,‘,˜|¨\Ó:€KØ×'Ñ'¨¨1Ó-×8Ñ8Ó:€Kà˜Ð$Ð$r1   Úpositions_idsÚbetaÚmax_position_embeddingsc           	      ó„   — d|t        j                  dt        j                  | |z  «      z   «      z  z   }|d d …d d d …d f   S )Nr"   )r*   ÚlogÚfloor)ra   rb   rc   rM   s       r/   Úget_llama_4_attn_scalerg   i   sB   € Ø�$œŸ™ 1¤u§{¡{°=ÐCZÑ3ZÓ'[Ñ#[Ó\Ñ\Ñ\€GØ’1�dšA˜tÐ#Ñ$Ð$r1   c                   ó  ‡ — 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j                  d
e
dz  dee   de	ej                  ej                  dz  f   fd„Zˆ xZS )ÚMinistral3Attentionz=Multi-headed attention from 'Attention Is All You Need' paperÚconfigÚ	layer_idxc                 ó  •— t         ‰| �  «        || _        || _        t	        |dd «      xs |j
                  |j                  z  | _        |j                  |j                  z  | _	        | j                  dz  | _
        |j                  | _        d| _        t        j                  |j
                  |j                  | j                  z  d¬«      | _        t        j                  |j
                  |j                  | j                  z  d¬«      | _        t        j                  |j
                  |j                  | j                  z  d¬«      | _        t        j                  |j                  | j                  z  |j
                  d¬«      | _        y )NrF   g      à¿TF©Úbias)ÚsuperÚ__init__rj   rk   ÚgetattrÚhidden_sizeÚnum_attention_headsrF   rD   rT   rM   Úattention_dropoutÚ	is_causalr   ÚLinearÚq_projÚk_projÚv_projÚo_proj©Úselfrj   rk   Ú	__class__s      €r/   rp   zMinistral3Attention.__init__r   s2  ø€ Ü‰ÑÔØˆŒØ"ˆŒÜ ¨
°DÓ9Òm¸V×=OÑ=OÐSY×SmÑSmÑ=mˆŒØ$*×$>Ñ$>À&×B\ÑB\Ñ$\ˆÔ!Ø—}‘} dÑ*ˆŒØ!'×!9Ñ!9ˆÔØˆŒÜ—i‘i × 2Ñ 2°F×4NÑ4NÐQU×Q^ÑQ^Ñ4^ÐejÔkˆŒÜ—i‘i × 2Ñ 2°F×4NÑ4NÐQU×Q^ÑQ^Ñ4^ÐejÔkˆŒÜ—i‘i × 2Ñ 2°F×4NÑ4NÐQU×Q^ÑQ^Ñ4^ÐejÔkˆŒÜ—i‘i × :Ñ :¸T¿]¹]Ñ JÈF×L^ÑL^ÐejÔkˆ�r1   Nr=   Úposition_embeddingsrL   Úposition_idsÚpast_key_valuesrO   r?   c           
      ó  — |j                   d d }g |¢d‘| j                  ‘­}| j                  |«      j                  |«      j	                  dd«      }	| j                  |«      j                  |«      j	                  dd«      }
| j                  |«      j                  |«      j	                  dd«      }|\  }}t        |	|
||«      \  }	}
|	t        || j                  j                  j                  d«      | j                  j                  j                  d«      «      j                  |	j                  «      z  }	|� |j                  |
|| j                  «      \  }
}t!        j"                  | j                  j$                  t&        «      } || |	|
||f| j(                  sdn| j*                  | j,                  t/        | j                  dd «      dœ|¤Ž\  }} |j0                  g |¢d‘­Ž j3                  «       }| j5                  |«      }||fS )	Nr%   r"   r&   Úllama_4_scaling_betaÚ original_max_position_embeddingsç        Úsliding_window)rN   rM   r…   )r)   rF   rw   ÚviewrV   rx   ry   r<   rg   rj   Úrope_parametersÚgetrZ   rQ   Úupdaterk   r   Úget_interfaceÚ_attn_implementationr`   rS   rt   rM   rq   rB   r[   rz   )r|   r=   r~   rL   r   r€   rO   Úinput_shapeÚhidden_shapeÚquery_statesr\   r]   r7   r8   Úattention_interfacer_   r^   s                    r/   ÚforwardzMinistral3Attention.forward€   sý  € ð $×)Ñ)¨#¨2Ð.ˆØ8˜Ð8 bÐ8¨$¯-©-Ñ8ˆà—{‘{ =Ó1×6Ñ6°|ÓD×NÑNÈqÐRSÓTˆØ—[‘[ Ó/×4Ñ4°\ÓB×LÑLÈQÐPQÓRˆ
Ø—{‘{ =Ó1×6Ñ6°|ÓD×NÑNÈqÐRSÓTˆà&‰ˆˆSÜ#7¸ÀjÐRUÐWZÓ#[Ñ ˆ�jØ#Ô&<ØØ�K‰K×'Ñ'×+Ñ+Ð,BÓCØ�K‰K×'Ñ'×+Ñ+Ð,NÓOó'
÷ ‰"ˆ\×ÑÓ
 ñ	!ˆð Ð&Ø'6×'=Ñ'=¸jÈ,ÐX\×XfÑXfÓ'gÑ$ˆJ˜ä(?×(MÑ(MØ�K‰K×,Ñ,Ô.Eó)
Ðñ %8ØØØØØð
%
ð  $Ÿ}š}‘C°$×2HÑ2HØ—L‘LÜ" 4§;¡;Ð0@À$ÓGñ
%
ð ñ
%
Ñ!ˆ�\ð *�k×)Ñ)Ð;¨;Ð;¸Ò;×FÑFÓHˆØ—k‘k +Ó.ˆØ˜LÐ(Ð(r1   ©N)Ú__name__Ú
__module__Ú__qualname__Ú__doc__r#   Úintrp   r*   ÚTensorÚtupler   r   r   r�   Ú__classcell__©r}   s   @r/   ri   ri   n   s¯   ø„ áGðlÐ/ð l¸Cõ lð( )-ñ-)à—|‘|ð-)ð # 5§<¡<°·±Ð#=Ñ>ð-)ð Ÿ™ tÑ+ð	-)ð
 —l‘lð-)ð  ™ð-)ð Ð-Ñ.ð-)ð 
ˆu�|‰|˜UŸ\™\¨DÑ0Ð0Ñ	1÷-)r1   ri   c                   ó$   ‡ — e Zd Zˆ fd„Zd„ Zˆ xZS )ÚMinistral3MLPc                 óÊ  •— t         ‰| �  «        || _        |j                  | _        |j                  | _        t        j                  | j                  | j                  d¬«      | _        t        j                  | j                  | j                  d¬«      | _        t        j                  | j                  | j                  d¬«      | _	        t        |j                     | _        y ©NFrm   )ro   rp   rj   rr   Úintermediate_sizer   rv   Ú	gate_projÚup_projÚ	down_projr   Ú
hidden_actÚact_fn©r|   rj   r}   s     €r/   rp   zMinistral3MLP.__init__±   s¦   ø€ Ü‰ÑÔØˆŒØ!×-Ñ-ˆÔØ!'×!9Ñ!9ˆÔÜŸ™ 4×#3Ñ#3°T×5KÑ5KÐRWÔXˆŒÜ—y‘y ×!1Ñ!1°4×3IÑ3IÐPUÔVˆŒÜŸ™ 4×#9Ñ#9¸4×;KÑ;KÐRWÔXˆŒÜ˜V×.Ñ.Ñ/ˆ�r1   c                 óˆ   — | j                  | j                  | j                  |«      «      | j                  |«      z  «      }|S r‘   )r¢   r¤   r    r¡   )r|   r,   r¢   s      r/   r�   zMinistral3MLP.forward»   s6   € Ø—N‘N 4§;¡;¨t¯~©~¸aÓ/@Ó#AÀDÇLÁLÐQRÃOÑ#SÓTˆ	ØÐr1   )r’   r“   r”   rp   r�   r™   rš   s   @r/   rœ   rœ   °   s   ø„ ô0ör1   rœ   ÚRMSNormc                   óh   ‡ — 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Ministral3RMSNormÚepsr?   Nc                 óŠ   •— t         ‰| �  «        t        j                  t	        j
                  |«      «      | _        || _        y)z@
        Ministral3RMSNorm is equivalent to T5LayerNorm
        N)ro   rp   r   Ú	Parameterr*   ÚonesÚweightÚvariance_epsilon)r|   rr   rª   r}   s      €r/   rp   zMinistral3RMSNorm.__init__Â   s1   ø€ ô 	‰ÑÔÜ—l‘l¤5§:¡:¨kÓ#:Ó;ˆŒØ #ˆÕr1   r=   c                 ó"  — |j                   }|j                  t        j                  «      }|j	                  d«      j                  dd¬«      }|t        j                  || j                  z   «      z  }| j                  |j                  |«      z  S )Nr&   r%   T)Úkeepdim)	rQ   rZ   r*   rY   ÚpowÚmeanÚrsqrtr¯   r®   )r|   r=   Úinput_dtypeÚvariances       r/   r�   zMinistral3RMSNorm.forwardÊ   sy   € Ø#×)Ñ)ˆØ%×(Ñ(¬¯©Ó7ˆØ ×$Ñ$ QÓ'×,Ñ,¨R¸Ð,Ó>ˆØ%¬¯©°H¸t×?TÑ?TÑ4TÓ(UÑUˆØ�{‰{˜]×-Ñ-¨kÓ:Ñ:Ð:r1   c                 ó^   — t        | j                  j                  «      › d| j                  › �S )Nz, eps=)r˜   r®   r)   r¯   )r|   s    r/   Ú
extra_reprzMinistral3RMSNorm.extra_reprÑ   s*   € Ü˜Ÿ™×)Ñ)Ó*Ð+¨6°$×2GÑ2GÐ1HÐIÐIr1   )g�íµ ÷Æ°>)
r’   r“   r”   Úfloatrp   r*   r—   r�   r¸   r™   rš   s   @r/   r©   r©   À   s7   ø„ ñ$¨ð $¸$õ $ð; U§\¡\ð ;°e·l±ló ;öJr1   r©   c                   ó  ‡ — e Zd Zdedefˆ fd„Z	 	 	 	 	 ddej                  dej                  dz  dej                  dz  de	dz  d	e
dz  d
eej                  ej                  f   dz  dee   dej                  fd„Zˆ xZS )ÚMinistral3DecoderLayerrj   rk   c                 ó   •— t         ‰| �  «        |j                  | _        t        ||¬«      | _        t        |«      | _        t        |j                  |j                  ¬«      | _	        t        |j                  |j                  ¬«      | _
        y )N)rj   rk   ©rª   )ro   rp   rr   ri   Ú	self_attnrœ   Úmlpr©   Úrms_norm_epsÚinput_layernormÚpost_attention_layernormr{   s      €r/   rp   zMinistral3DecoderLayer.__init__Ö   sm   ø€ Ü‰ÑÔØ!×-Ñ-ˆÔÜ,°FÀiÔPˆŒÜ  Ó(ˆŒÜ0°×1CÑ1CÈ×I\ÑI\Ô]ˆÔÜ(9¸&×:LÑ:LÐRX×ReÑReÔ(fˆÕ%r1   Nr=   rL   r   r€   Ú	use_cacher~   rO   r?   c           
      óÀ   — |}| j                  |«      } | j                  d||||||dœ|¤Ž\  }}	||z   }|}| j                  |«      }| j                  |«      }||z   }|S )N)r=   rL   r   r€   rÃ   r~   © )rÁ   r¾   rÂ   r¿   )
r|   r=   rL   r   r€   rÃ   r~   rO   ÚresidualÚ_s
             r/   r�   zMinistral3DecoderLayer.forwardÞ   s�   € ð !ˆØ×,Ñ,¨]Ó;ˆà)˜4Ÿ>™>ð 
Ø'Ø)Ø%Ø+ØØ 3ñ
ð ñ
Ñˆ�qð ! =Ñ0ˆð !ˆØ×5Ñ5°mÓDˆØŸ™ Ó/ˆØ  =Ñ0ˆØÐr1   )NNNFN)r’   r“   r”   r#   r–   rp   r*   r—   Ú
LongTensorr   Úboolr˜   r   r   r�   r™   rš   s   @r/   r»   r»   Õ   s¿   ø„ ðgÐ/ð g¸Cõ gð /3Ø04Ø(,Ø!&ØHLñà—|‘|ðð Ÿ™ tÑ+ðð ×&Ñ&¨Ñ-ð	ð
  ™ðð ˜$‘;ðð # 5§<¡<°·±Ð#=Ñ>ÀÑEðð Ð+Ñ,ðð 
�‰÷r1   r»   c                   óJ   — 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œZy)ÚMinistral3PreTrainedModelrj   ÚmodelTr»   r€   )r=   Ú
attentionsN)r’   r“   r”   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»   ri   Ú_can_record_outputsrÅ   r1   r/   rË   rË   þ   sQ   … àÓØÐØ&*Ð#Ø1Ð2ÐØ#4Ð"5ÐØÐØ€NØÐà!ÐØ"&Ðà/Ø)ñÑr1   rË   c                   óÄ   ‡ — 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Ministral3RotaryEmbeddingÚinv_freqNrj   c                 ó®  •— t         ‰| �  «        |j                  | _        |j                  | _        || _        | j
                  j                  d   | _        | j                  }| j                  dk7  rt        | j                     } || j
                  |«      \  }| _
        | j                  d|d¬«       | j                  d|j                  «       d¬«       y )NÚ	rope_typeÚdefaultrÛ   F)Ú
persistentÚoriginal_inv_freq)ro   rp   rc   Úmax_seq_len_cachedÚoriginal_max_seq_lenrj   r‡   rÝ   Úcompute_default_rope_parametersr   Úattention_scalingÚregister_bufferÚclone)r|   rj   ÚdeviceÚrope_init_fnrÛ   r}   s        €r/   rp   z"Ministral3RotaryEmbedding.__init__  s¶   ø€ Ü‰ÑÔØ"(×"@Ñ"@ˆÔØ$*×$BÑ$BˆÔ!àˆŒàŸ™×4Ñ4°[ÑAˆŒØ!%×!EÑ!EˆØ�>‰>˜YÒ&Ü.¨t¯~©~Ñ>ˆLÙ+7¸¿¹ÀVÓ+LÑ(ˆ�$Ô(à×Ñ˜Z¨¸eÐÔDØ×ÑÐ0°(·.±.Ó2BÈuÐÕUr1   rç   ztorch.deviceÚseq_lenr?   ztorch.Tensorc                 ó  — | j                   d   }t        | dd«      xs | j                  | j                  z  }d}d|t	        j
                  d|dt        j                  ¬«      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_thetarF   Ng      ð?r   r&   ©rQ   )rç   rQ   )	r‡   rq   rr   rs   r*   ÚarangeÚint64rZ   r¹   )rj   rç   ré   Úbaser(   Úattention_factorrÛ   s          r/   rã   z9Ministral3RotaryEmbedding.compute_default_rope_parameters$  s�   € ð& ×%Ñ% lÑ3ˆÜ�f˜j¨$Ó/Òc°6×3EÑ3EÈ×IcÑIcÑ3cˆàÐð Ø”U—\‘\ ! S¨!´5·;±;Ô?×BÑBÈ&ÔX]×XcÑXcÐBÓdÐgjÑjÑkñ
ˆð Ð)Ð)Ð)r1   c                 óN  — | j                   d d d …d f   j                  «       j                  |j                  d   dd«      j	                  |j
                  «      }|d d …d d d …f   j                  «       }t        |j
                  j                  t        «      r/|j
                  j                  dk7  r|j
                  j                  nd}t        |d¬«      5  |j                  «       |j                  «       z  j                  dd«      }t        j                  ||fd¬	«      }|j                  «       | j                  z  }|j                  «       | j                  z  }	d d d «       j	                  |j                   ¬
«      	j	                  |j                   ¬
«      fS # 1 sw Y   ŒAxY w)Nr   r%   r"   ÚmpsÚcpuF)Údevice_typeÚenabledr&   r'   rì   )rÛ   r¹   rA   r)   rZ   rç   Ú
isinstanceÚtypeÚstrr   rV   r*   r+   r7   rä   r8   rQ   )
r|   r,   r   Úinv_freq_expandedÚposition_ids_expandedrô   ÚfreqsÚembr7   r8   s
             r/   r�   z!Ministral3RotaryEmbedding.forwardB  sR  € ð !ŸM™M¨$²°4¨-Ñ8×>Ñ>Ó@×GÑGÈ×HZÑHZÐ[\ÑH]Ð_aÐcdÓe×hÑhÐij×iqÑiqÓrÐØ ,ªQ°²a¨ZÑ 8× >Ñ >Ó @Ðä'1°!·(±(·-±-ÄÔ'EÈ!Ï(É(Ï-É-Ð[`ÒJ`�a—h‘h—m’mÐfkˆÜ¨¸UÔCñ 	5Ø&×,Ñ,Ó.Ð1F×1LÑ1LÓ1NÑN×YÑYÐZ[Ð]^Ó_ˆEÜ—)‘)˜U E˜N°Ô3ˆCØ—'‘'“)˜d×4Ñ4Ñ4ˆCØ—'‘'“)˜d×4Ñ4Ñ4ˆC÷		5ð �v‰v˜AŸG™GˆvÓ$ c§f¡f°1·7±7 fÓ&;Ð;Ð;÷	5ð 	5ús   ÃBFÆF$r‘   )NNN)r’   r“   r”   r*   r—   rÎ   r#   rp   Ústaticmethodr   r–   r˜   r¹   rã   Úno_gradr   r�   r™   rš   s   @r/   rÚ   rÚ     s™   ø… Ø�l‰lÓñVÐ/õ Vð  à*.Ø+/Ø"ñ*Ø  4Ñ'ð*à˜Ñ(ð*ð �t‘ð*ð 
ˆ~˜uÐ$Ñ	%ò	*ó ð*ð: €U‡]�]ƒ_Øñ<ó ó ô<r1   rÚ   c                   óò   ‡ — 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Ministral3Modelrj   c           	      ó  •— t         ‰| �  |«       |j                  | _        |j                  | _        t        j                  |j                  |j                  | j                  «      | _        t        j                  t        |j                  «      D �cg c]  }t        ||«      ‘Œ c}«      | _        t        |j                  |j                  ¬«      | _        t#        |¬«      | _        d| _        | j)                  «        y c c}w )Nr½   ©rj   F)ro   rp   Úpad_token_idÚpadding_idxÚ
vocab_sizer   Ú	Embeddingrr   Úembed_tokensÚ
ModuleListÚrangeÚnum_hidden_layersr»   Úlayersr©   rÀ   ÚnormrÚ   Ú
rotary_embÚgradient_checkpointingÚ	post_initr{   s      €r/   rp   zMinistral3Model.__init__T  sÆ   ø€ Ü‰Ñ˜Ô Ø!×.Ñ.ˆÔØ ×+Ñ+ˆŒäŸL™L¨×):Ñ):¸F×<NÑ<NÐPT×P`ÑP`ÓaˆÔÜ—m‘mÜHMÈf×NfÑNfÓHgÖh¸9Ô# F¨IÕ6Òhó
ˆŒô & f×&8Ñ&8¸f×>QÑ>QÔRˆŒ	Ü3¸6ÔBˆŒØ&+ˆÔ#ð 	�‰Õùò is   ÂDNÚ	input_idsrL   r   r€   Úinputs_embedsrÃ   rO   r?   c           
      óŒ  — |d u |d uz  rt        d«      ‚|€| j                  |«      }|r|€t        | j                  ¬«      }|€V|�|j	                  «       nd}t        j                  |j                  d   |j                  ¬«      |z   }|j                  d«      }| j                  j                  €t        nt        }	 |	| j                  ||||¬«      }
|}| j                  ||¬«      }| j                  d | j                  j                   D ]  } ||f|
||||dœ|¤Ž}Œ | j!                  |«      }t#        ||r|¬	«      S d ¬	«      S )
Nz:You must specify exactly one of input_ids or inputs_embedsr  r   r"   )rç   )rj   r  rL   r€   r   )r   )rL   r   r€   rÃ   r~   )Úlast_hidden_stater€   )Ú
ValueErrorr  r	   rj   Úget_seq_lengthr*   rí   r)   rç   r4   r…   r   r   r  r  r
  r  r   )r|   r  rL   r   r€   r  rÃ   rO   Úpast_seen_tokensÚmask_functionÚcausal_maskr=   r~   Údecoder_layers                 r/   r�   zMinistral3Model.forwardd  ss  € ð ˜Ð -°tÐ";Ò<ÜÐYÓZÐZàÐ Ø ×-Ñ-¨iÓ8ˆMá˜Ð0Ü*°$·+±+Ô>ˆOàÐØCRÐC^˜×=Ñ=Ô?ÐdeÐÜ Ÿ<™<¨×(;Ñ(;¸AÑ(>À}×G[ÑG[Ô\Ð_oÑoˆLØ'×1Ñ1°!Ó4ˆLà.2¯k©k×.HÑ.HÐ.PÕ*ÔVwˆÙ#Ø—;‘;Ø'Ø)Ø+Ø%ô
ˆð &ˆØ"Ÿo™o¨mÈ,˜oÓWÐà!Ÿ[™[Ð)H¨4¯;©;×+HÑ+HÐIò 		ˆMÙ)Øðà*Ø)Ø /Ø#Ø$7ñð ñ‰Mð		ð Ÿ	™	 -Ó0ˆÜ&Ø+Ù/8˜Oô
ð 	
à>Bô
ð 	
r1   )NNNNNN)r’   r“   r”   r#   rp   r    r!   r   r*   rÈ   r—   r   ÚFloatTensorrÉ   r   r   r   r�   r™   rš   s   @r/   r   r   R  sÎ   ø„ ðÐ/õ ð   ØØð .2Ø.2Ø04Ø(,Ø26Ø!%ñ2
à×#Ñ# dÑ*ð2
ð Ÿ™ tÑ+ð2
ð ×&Ñ&¨Ñ-ð	2
ð
  ™ð2
ð ×(Ñ(¨4Ñ/ð2
ð ˜$‘;ð2
ð Ð+Ñ,ð2
ð 
!ò2
ó ó ó  ô2
r1   r   c                   óB  ‡ — 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e   defd„«       «       Zˆ xZS )ÚMinistral3ForCausalLMzlm_head.weightzmodel.embed_tokens.weightÚlm_headÚcolwise_gather_outputr=   Úlogitsc                 óè   •— t         ‰| �  |«       t        |«      | _        |j                  | _        t        j                  |j                  |j                  d¬«      | _        | j                  «        y rž   )
ro   rp   r   rÌ   r  r   rv   rr   r  r  r¥   s     €r/   rp   zMinistral3ForCausalLM.__init__¢  sU   ø€ Ü‰Ñ˜Ô Ü$ VÓ,ˆŒ
Ø ×+Ñ+ˆŒÜ—y‘y ×!3Ñ!3°V×5FÑ5FÈUÔSˆŒð 	�‰Õr1   Nr  rL   r   r€   r  ÚlabelsrÃ   Úlogits_to_keeprO   r?   c	           
      óx  —  | j                   d||||||dœ|	¤Ž}
|
j                  }t        |t        «      rt	        | d«      n|}| j                  |dd…|dd…f   «      }d}|�* | j                  d||| j                  j                  dœ|	¤Ž}t        |||
j                  |
j                  |
j                  ¬«      S )aí  
        Example:

        ```python
        >>> from transformers import AutoTokenizer, Ministral3ForCausalLM

        >>> model = Ministral3ForCausalLM.from_pretrained("meta-ministral3/Ministral3-2-7b-hf")
        >>> tokenizer = AutoTokenizer.from_pretrained("meta-ministral3/Ministral3-2-7b-hf")

        >>> 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."
        ```)r  rL   r   r€   r  rÃ   N)r  r!  r  )Úlossr  r€   r=   rÍ   rÅ   )rÌ   r  rö   r–   Úslicer  Úloss_functionrj   r  r   r€   r=   rÍ   )r|   r  rL   r   r€   r  r!  rÃ   r"  rO   Úoutputsr=   Úslice_indicesr  r$  s                  r/   r�   zMinistral3ForCausalLM.forward«  sÙ   € ð> ,6¨4¯:©:ð ,
ØØ)Ø%Ø+Ø'Øñ,
ð ñ,
ˆð  ×1Ñ1ˆä8BÀ>ÔSVÔ8Wœ˜~˜o¨tÔ4Ð]kˆØ—‘˜mªA¨}ºaÐ,?Ñ@ÓAˆàˆØÐØ%�4×%Ñ%Ðp¨V¸FÈtÏ{É{×OeÑOeÑpÐioÑpˆDä%ØØØ#×3Ñ3Ø!×/Ñ/Ø×)Ñ)ô
ð 	
r1   )NNNNNNNr   )r’   r“   r”   Ú_tied_weights_keysÚ_tp_planÚ_pp_planrp   r   r   r*   rÈ   r—   r   r  rÉ   r–   r   r   r   r�   r™   rš   s   @r/   r  r  œ  s  ø„ à*Ð,GÐHÐØÐ2Ð3€HØ˜_Ð-°¨zÐ:Ð;€Hôð Øð .2Ø.2Ø04Ø(,Ø26Ø*.Ø!%Ø-.ñ6
à×#Ñ# dÑ*ð6
ð Ÿ™ tÑ+ð6
ð ×&Ñ&¨Ñ-ð	6
ð
  ™ð6
ð ×(Ñ(¨4Ñ/ð6
ð × Ñ  4Ñ'ð6
ð ˜$‘;ð6
ð ˜eŸl™lÑ*ð6
ð Ð+Ñ,ð6
ð 
 ò6
ó ó ô6
r1   r  c                   ó   — e Zd Zy)Ú Ministral3ForTokenClassificationN©r’   r“   r”   rÅ   r1   r/   r-  r-  æ  ó   „ Ør1   r-  c                   ó   — e Zd Zy)Ú#Ministral3ForSequenceClassificationNr.  rÅ   r1   r/   r1  r1  ê  r/  r1   r1  c                   ó   — e Zd Zy)ÚMinistral3ForQuestionAnsweringNr.  rÅ   r1   r/   r3  r3  î  r/  r1   r3  )r  r3  r   rË   r1  r-  )r"   )r„   )FÚcollections.abcr   Útypingr   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   r   r   r   Úmodeling_outputsr   r   Úmodeling_rope_utilsr   r   Úmodeling_utilsr   r   Úprocessing_utilsr   Úutilsr   r   r   Úutils.genericr   r    Úutils.output_capturingr!   Úconfiguration_ministral3r#   r0   r<   r—   r–   rG   ÚModuler¹   r`   rg   ri   rœ   r©   r»   rË   rÚ   r   r  r-  r1  r3  Ú__all__rÅ   r1   r/   ú<module>rG     sG  ðõ %Ý ã Ý å !ß .Ý )ß fÑ fß RÝ B÷ó ÷ Pß Kß FÝ &ß IÑ Iß GÝ 5Ý 6ò(ñ Ð*Ó+òó ,ðð2	U˜UŸ\™\ð 	U°#ð 	U¸%¿,¹,ó 	Uð& ñ%Ø�I‰Ið%à�<‰<ð%ð 
�‰ð%ð �<‰<ð	%ð
 —L‘L 4Ñ'ð%ð ð%ð ð%ð Ð'Ñ(ó%ð2%¨%¯,©,ð %¸eð %Ð^að %Ðfk×frÑfró %ñ
 Ð)Ó*ô>)˜"Ÿ)™)ó >)ó +ð>)ôB�B—I‘Iô ñ  ˜YÓ'ôJ˜Ÿ	™	ó Jó (ðJô(&Ð7ô &ðR ô ó ó ðô$>< §	¡	ô ><ðB ôF
Ð/ó F
ó ðF
ðR ôF
Ð5°ó F
ó ðF
ôR	Ð'DÐF_ô 	ô	Ð*JÐLeô 	ô	Ð%@ÐB[ô 	ò�r1   