
    iW                        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)D    )Callable)OptionalN)nn   )ACT2FN)CacheDynamicCache)GenerationMixin)use_kernel_forward_from_hubuse_kernel_func_from_hubuse_kernelized_func)create_causal_mask!create_sliding_window_causal_mask)FlashAttentionKwargs)GenericForQuestionAnswering GenericForSequenceClassificationGenericForTokenClassificationGradientCheckpointingLayer)BaseModelOutputWithPastCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstringcan_return_tuple)maybe_autocastmerge_with_config_defaults)capture_outputs   )Ministral3Configc                     | 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..N   dim)shapetorchcat)xx1x2s      n/mnt/e/genesis-system/.venv/lib/python3.12/site-packages/transformers/models/ministral3/modeling_ministral3.pyrotate_halfr0   #   sZ    	
3"!''"+"""	#B	
3q ""	#B99rc2YB''    rotary_pos_embc                     |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.
    )	unsqueezer0   )qkcossinunsqueeze_dimq_embedk_embeds          r/   apply_rotary_pos_embr<   *   sY    & --
&C
--
&C3w;q>C/0G3w;q>C/0GGr1   hidden_statesn_repreturnc                     | j                   \  }}}}|dk(  r| S | dddddddddf   j                  |||||      } | j                  |||z  ||      S )z
    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)   expandreshape)r=   r>   batchnum_key_value_headsslenhead_dims         r/   	repeat_kvrG   D   so    
 2?1D1D.Ehz!!Qa"23::5BUW\^bdlmM  (;e(CT8TTr1   modulequerykeyvalueattention_maskscalingdropoutkwargsc                    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(   dtype)ptrainingr"   )rG   num_key_value_groupsr*   matmul	transposer   
functionalsoftmaxfloat32torQ   rN   rS   
contiguous)rH   rI   rJ   rK   rL   rM   rN   rO   
key_statesvalue_statesattn_weightsattn_outputs               r/   eager_attention_forwardr`   P   s     3 ; ;<JUF$?$?@L<<z';';Aq'ABWLL!#n4==((2U]](SVVW\WbWbcL==((6??([L,,|\:K''1-88:K$$r1   positions_idsbetamax_position_embeddingsc           	          d|t        j                  dt        j                  | |z        z         z  z   }|j                  d      S )Nr"   r%   )r*   logfloorr4   )ra   rb   rc   rM   s       r/   _get_llama_4_attn_scalerg   i   s?    $1u{{=CZ3Z'[#[\\\GR  r1   c                       e Zd ZdZdedef 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j                  dz  dee   de	ej                  ej                  dz  f   fdZ xZS )Ministral3Attentionz=Multi-headed attention from 'Attention Is All You Need' paperconfig	layer_idxc                    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      TFbias)super__init__rj   rk   getattrhidden_sizenum_attention_headsrF   rD   rT   rM   attention_dropout	is_causalr   Linearq_projk_projv_projo_projselfrj   rk   	__class__s      r/   rp   zMinistral3Attention.__init__r   s2   "
D9mV=O=OSYSmSm=m$*$>$>&B\B\$\!}}d*!'!9!9ii 2 2F4N4NQUQ^Q^4^ejkii 2 2F4N4NQUQ^Q^4^ejkii 2 2F4N4NQUQ^Q^4^ejkii : :T]] JFL^L^ejkr1   Nr=   position_embeddingsrL   past_key_valuescache_positionrO   r?   c           
      "   |j                   d d }g |d| j                  }| j                  |      j                  |      j	                  dd      }	| j                  |      j                  |      j	                  dd      }
| j                  |      j                  |      j	                  dd      }|\  }}t        |	|
||      \  }	}
|	t        || j                  j                  j                  d      | j                  j                  j                  d            j                  |	j                        z  }	|'|||d}|j                  |
|| j                  |      \  }
}t!        j"                  | j                  j$                  t&              } || |	|
||f| j(                  sdn| j*                  | j,                  t/        | j                  dd       d	|\  }} |j0                  g |d j3                         }| j5                  |      }||fS )
Nr%   r"   r&   llama_4_scaling_beta original_max_position_embeddings)r8   r7   r           sliding_window)rN   rM   r   )r)   rF   rw   viewrV   rx   ry   r<   rg   rj   rope_parametersgetrZ   rQ   updaterk   r   get_interface_attn_implementationr`   rS   rt   rM   rq   rB   r[   rz   )r|   r=   r~   rL   r   r   rO   input_shapehidden_shapequery_statesr\   r]   r7   r8   cache_kwargsattention_interfacer_   r^   s                     r/   forwardzMinistral3Attention.forward   s    $))#2.88b8$--8{{=166|DNNqRST[[/44\BLLQPQR
{{=166|DNNqRST&S#7jRUWZ#[ j#&=KK''++,BCKK''++,NO'
 "\
 	! &#&snUL'6'='=j,X\XfXfht'u$J(?(M(MKK,,.E)
 %8
%
  $}}C$2H2HLL"4;;0@$G
%
 
%
!\ *k));;;;FFHkk+.L((r1   )NN)__name__
__module____qualname____doc__r#   intrp   r*   Tensortupler   
LongTensorr   r   r   __classcell__r}   s   @r/   ri   ri   n   s    Gl/ lC l& )-26/)||/) #5<<#=>/) t+	/)
 /) ((4//) -./) 
u||U\\D00	1/)r1   ri   c                   $     e Zd Z fdZd Z xZS )Ministral3MLPc                    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   intermediate_sizer   rv   	gate_projup_proj	down_projr   
hidden_actact_fnr|   rj   r}   s     r/   rp   zMinistral3MLP.__init__   s    !--!'!9!94#3#3T5K5KRWXyy!1!143I3IPUV4#9#94;K;KRWXV../r1   c                     | j                  | j                  | j                  |            | j                  |      z        }|S N)r   r   r   r   )r|   r,   r   s      r/   r   zMinistral3MLP.forward   s6    NN4;;t~~a/@#ADLLQRO#ST	r1   )r   r   r   rp   r   r   r   s   @r/   r   r      s    0r1   r   RMSNormc                   h     e Zd Zddeddf fdZdej                  dej                  fdZd Z xZ	S )	Ministral3RMSNormepsr?   Nc                     t         |           t        j                  t	        j
                  |            | _        || _        y)z@
        Ministral3RMSNorm is equivalent to T5LayerNorm
        N)ro   rp   r   	Parameterr*   onesweightvariance_epsilon)r|   rr   r   r}   s      r/   rp   zMinistral3RMSNorm.__init__   s1     	ll5::k#:; #r1   r=   c                 "   |j                   }|j                  t        j                        }|j	                  d      j                  dd      }|t        j                  || j                  z         z  }| j                  |j                  |      z  S )Nr&   r%   T)keepdim)	rQ   rZ   r*   rY   powmeanrsqrtr   r   )r|   r=   input_dtypevariances       r/   r   zMinistral3RMSNorm.forward   sy    #))%((7 $$Q',,R,>%Ht?T?T4T(UU{{]--k:::r1   c                 ^    t        | j                  j                         d| j                   S )Nz, eps=)r   r   r)   r   )r|   s    r/   
extra_reprzMinistral3RMSNorm.extra_repr   s*    ))*+6$2G2G1HIIr1   )gư>)
r   r   r   floatrp   r*   r   r   r   r   r   s   @r/   r   r      s7    $ $$ $;U\\ ;ell ;Jr1   r   c                   "    e Zd Zdedef fdZ	 	 	 	 	 	 ddej                  dej                  dz  dej                  dz  de	dz  d	e
dz  d
ej                  dz  deej                  ej                  f   dz  dee   dej                  fdZ xZS )Ministral3DecoderLayerrj   rk   c                     t         |           |j                  | _        t        ||      | _        t        |      | _        t        |j                  |j                        | _	        t        |j                  |j                        | _
        y )N)rj   rk   r   )ro   rp   rr   ri   	self_attnr   mlpr   rms_norm_epsinput_layernormpost_attention_layernormr{   s      r/   rp   zMinistral3DecoderLayer.__init__   sm    !--,FiP (01C1CI\I\](9&:L:LRXReRe(f%r1   Nr=   rL   position_idsr   	use_cacher   r~   rO   r?   c                     |}	| j                  |      } | j                  d|||||||d|\  }}
|	|z   }|}	| j                  |      }| j                  |      }|	|z   }|S )N)r=   rL   r   r   r   r   r~    )r   r   r   r   )r|   r=   rL   r   r   r   r   r~   rO   residual_s              r/   r   zMinistral3DecoderLayer.forward   s     !,,];)4>> 	
')%+) 3	
 	
q !=0 !55mD/ =0r1   )NNNFNN)r   r   r   r#   r   rp   r*   r   r   r   boolr   r   r   r   r   r   s   @r/   r   r      s    g/ gC g /304(,!&26HL|| t+ &&-	
  $; ((4/ #5<<#=>E +, 
r1   r   c                   J    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Zy)Ministral3PreTrainedModelrj   modelTr   r   )r=   
attentionsN)r   r   r   r#   __annotations__base_model_prefixsupports_gradient_checkpointing_no_split_modules_skip_keys_device_placement_supports_flash_attn_supports_sdpa_supports_flex_attn_can_compile_fullgraph_supports_attention_backendr   ri   _can_record_outputsr   r1   r/   r   r     sQ    &*#12#4"5N!"&/)r1   r   c                        e Zd ZU ej                  ed<   ddef 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 xZS )Ministral3RotaryEmbeddinginv_freqNrj   c                    t         |           |j                  | _        |j                  | _        || _        | j
                  j                  d   | _        | j                  }| j                  dk7  rt        | j                     } || j
                  |      \  }| _
        | j                  d|d       | j                  d|j                         d       y )N	rope_typedefaultr   F)
persistentoriginal_inv_freq)ro   rp   rc   max_seq_len_cachedoriginal_max_seq_lenrj   r   r   compute_default_rope_parametersr   attention_scalingregister_bufferclone)r|   rj   devicerope_init_fnr   r}   s        r/   rp   z"Ministral3RotaryEmbedding.__init__  s    "("@"@$*$B$B!44[A!%!E!E>>Y&.t~~>L+7V+L($(ZeD0(..2BuUr1   r   ztorch.deviceseq_lenr?   ztorch.Tensorc                    | 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).
        
rope_thetarF   Ng      ?r   r&   rQ   )r   rQ   )	r   rq   rr   rs   r*   arangeint64rZ   r   )rj   r   r   baser(   attention_factorr   s          r/   r   z9Ministral3RotaryEmbedding.compute_default_rope_parameters(  s    & %%l3fj$/c63E3EIcIc3c U\\!S!5;;?BB&X]XcXcBdgjjk
 )))r1   c                 N   | j                   d d d d f   j                         j                  |j                  d   dd      j	                  |j
                        }|d d d d d f   j                         }t        |j
                  j                  t              r/|j
                  j                  dk7  r|j
                  j                  nd}t        |d      5  |j                         |j                         z  j                  dd      }t        j                  ||fd	      }|j                         | j                  z  }|j                         | j                  z  }	d d d        j	                  |j                   
      	j	                  |j                   
      fS # 1 sw Y   AxY w)Nr   r%   r"   mpscpuF)device_typeenabledr&   r'   r   )r   r   rA   r)   rZ   r   
isinstancetypestrr   rV   r*   r+   r7   r   r8   rQ   )
r|   r,   r   inv_freq_expandedposition_ids_expandedr   freqsembr7   r8   s
             r/   r   z!Ministral3RotaryEmbedding.forwardF  sR    !MM$4-8>>@GGHZHZ[\H]_acdehhijiqiqr ,QaZ 8 > > @'1!((--'E!((--[`J`ahhmmfkUC 	5&,,.1F1L1L1NNYYZ[]^_E))UEN3C'')d444C'')d444C		5 vvAGGv$cff177f&;;;	5 	5s   BFF$r   )NNN)r   r   r   r*   r   r   r#   rp   staticmethodr   r   r   r   r   no_gradr   r   r   r   s   @r/   r   r     s    llV/ V  *.+/"* 4'*(* t* 
~u$	%	* *: U]]_<  <r1   r   c                       e Zd Zdef fdZeee	 	 	 	 	 	 	 ddej                  dz  dej                  dz  dej                  dz  dedz  dej                  dz  d	edz  d
ej                  dz  dee   defd                     Z xZS )Ministral3Modelrj   c           	         t         |   |       |j                  | _        |j                  | _        t        j                  |j                  |j                  | j                        | _        t        j                  t        |j                        D cg c]  }t        ||       c}      | _        t        |j                  |j                        | _        t#        |      | _        d| _        | j)                          y c c}w )Nr   rj   F)ro   rp   pad_token_idpadding_idx
vocab_sizer   	Embeddingrr   embed_tokens
ModuleListrangenum_hidden_layersr   layersr   r   normr   
rotary_embgradient_checkpointing	post_initr{   s      r/   rp   zMinistral3Model.__init__X  s     !.. ++LL):):F<N<NPTP`P`ammHMfNfNfHgh9#FI6h
 &f&8&8f>Q>QR	36B&+# 	 is   DN	input_idsrL   r   r   inputs_embedsr   r   rO   r?   c                    |d u |d uz  rt        d      || j                  |      }|r|t        | j                        }|F||j	                         nd}	t        j                  |	|	|j                  d   z   |j                        }||j                  d      }| j                  j                  t        nt        }
 |
| j                  |||||      }|}| j                  ||      }| j                  d | j                  j                   D ]  } ||f||||||d|} | j!                  |      }t#        ||r|	      S d 	      S )
Nz:You must specify exactly one of input_ids or inputs_embedsr  r   r"   )r   )rj   r  rL   r   r   r   )r   )rL   r   r   r   r   r~   )last_hidden_stater   )
ValueErrorr	  r	   rj   get_seq_lengthr*   r   r)   r   r4   r   r   r   r  r  r  r  r   )r|   r  rL   r   r   r  r   r   rO   past_seen_tokensmask_functioncausal_maskr=   r~   decoder_layers                  r/   r   zMinistral3Model.forwardh  s    -t";<YZZ  --i8M0*$++>O!CRC^==?de"\\ "2]5H5H5K"KTaThThN )33A6L.2kk.H.H.P*Vw#;;'))+%
 &"oom,oW![[)H4;;+H+HI 
	M)	*) /#-$7	 	M
	 		-0&+/8O
 	
>B
 	
r1   )NNNNNNN)r   r   r   r#   rp   r    r!   r   r*   r   r   r   FloatTensorr   r   r   r   r   r   r   s   @r/   r  r  V  s    /     .2.204(,26!%269
##d*9
 t+9
 &&-	9

 9
 ((4/9
 $;9
 ((4/9
 +,9
 
!9
    9
r1   r  c                   b    e Zd ZddiZddiZddgdgfiZ fdZee	 	 	 	 	 	 	 	 	 dd	e	j                  dz  d
e	j                  dz  de	j                  dz  dedz  de	j                  dz  de	j                  dz  dedz  de	j                  dz  dee	j                  z  dee   defd              Z xZS )Ministral3ForCausalLMzlm_head.weightzmodel.embed_tokens.weightlm_headcolwise_gather_outputr=   logitsc                     t         |   |       t        |      | _        |j                  | _        t        j                  |j                  |j                  d      | _        | j                          y r   )
ro   rp   r  r   r  r   rv   rr   r  r  r   s     r/   rp   zMinistral3ForCausalLM.__init__  sU     $V,
 ++yy!3!3V5F5FUS 	r1   Nr  rL   r   r   r  labelsr   r   logits_to_keeprO   r?   c
                 z    | j                   d|||||||d|
}|j                  }t        |	t              rt	        |	 d      n|	}| j                  |dd|ddf         }d}|* | j                  d||| j                  j                  d|
}t        |||j                  |j                  |j                        S )a  
        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  r   r   N)r!  r#  r  )lossr!  r   r=   r   r   )r   r  r   r   slicer  loss_functionrj   r  r   r   r=   r   )r|   r  rL   r   r   r  r#  r   r   r$  rO   outputsr=   slice_indicesr!  r&  s                   r/   r   zMinistral3ForCausalLM.forward  s    @ ,64:: 	,
)%+')	,
 	,
  118B>SV8W~ot4]kmA}a,?@A%4%%pVFt{{OeOepiopD%#33!//))
 	
r1   )	NNNNNNNNr   )r   r   r   _tied_weights_keys_tp_plan_pp_planrp   r   r   r*   r   r   r   r  r   r   r   r   r   r   r   r   s   @r/   r  r    s/   *,GH23H_-z:;H  .2.204(,26*.!%26-.8
##d*8
 t+8
 &&-	8

 8
 ((4/8
   4'8
 $;8
 ((4/8
 ell*8
 +,8
 
 8
  8
r1   r  c                       e Zd Zy) Ministral3ForTokenClassificationNr   r   r   r   r1   r/   r/  r/        r1   r/  c                       e Zd Zy)#Ministral3ForSequenceClassificationNr0  r   r1   r/   r3  r3    r1  r1   r3  c                       e Zd Zy)Ministral3ForQuestionAnsweringNr0  r   r1   r/   r5  r5    r1  r1   r5  )r  r5  r  r   r3  r/  )r"   )r   )Fcollections.abcr   typingr   r*   r   activationsr   cache_utilsr   r	   
generationr
   integrationsr   r   r   masking_utilsr   r   modeling_flash_attention_utilsr   modeling_layersr   r   r   r   modeling_outputsr   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   utils.genericr   r    utils.output_capturingr!   configuration_ministral3r#   r0   r<   r   r   rG   Moduler   r`   rg   ri   r   r   r   r   r   r  r  r/  r3  r5  __all__r   r1   r/   <module>rI     sJ   %    ! . ) f f R B  P K F & I I G 5 6( *+ ,2	UU\\ 	U# 	U%,, 	U& %II%<<% 
% <<	%
 LL4'% % % '(%2!5<< !u !_b !glgsgs !
 )*@)")) @) +@)FBII   Y'J		 J (J((7 (V   $><		 ><B M
/ M
 M
` H
5 H
 H
V	'DF_ 		*JLe 		%@B[ 	r1   