
    Z j'|                     n   S SK Jr  S SKJr  S SKrS SKJs  Jr  S SKJr  SSK	J
r  SSKJr  SSKJrJr  SS	KJr  SS
KJrJrJr  SSKJr  SSKJr  SSKJrJrJr  SSKJ r J!r!  SSK"J#r#J$r$  SSK%J&r&J'r'  SSK(J)r)  SSK*J+r+J,r,J-r-  SSK.J/r/J0r0J1r1  SSK2J3r3  SSK4J5r5  \" S5       " S S\Rl                  5      5       r7 " S S\Rl                  5      r8 " S S\Rl                  5      r9 " S S\Rl                  5      r:\ " S  S!\Rl                  5      5       r; " S" S#\Rl                  5      r<S$ r=\" S%5      SHS& j5       r>S'\R~                  S(\@S)\R~                  4S* jrA SIS+\Rl                  S,\R~                  S-\R~                  S.\R~                  S/\R~                  S-  S0\BS1\BS2\)\+   4S3 jjrCSJS4 jrDS5\R~                  S6\BS7\@S)\R~                  4S8 jrE " S9 S:\Rl                  5      rF " S; S<\5      rG " S= S>\'5      rH\, " S? S@\H5      5       rI\, " SA SB\H\5      5       rJ " SC SD\\H5      rK " SE SF\\H5      rL/ SGQrMg)K    )Callable)OptionalN)nn   )initialization)ACT2FN)CacheDynamicCache)GenerationMixin)use_experts_implementationuse_kernel_forward_from_hubuse_kernel_func_from_hub)create_causal_mask)FlashAttentionKwargs) GenericForSequenceClassificationGenericForTokenClassificationGradientCheckpointingLayer)BaseModelOutputWithPastCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstringcan_return_tuple)is_flash_attention_requestedmaybe_autocastmerge_with_config_defaults)capture_outputs   )Mistral4ConfigRMSNormc                   x   ^  \ rS rSrS
S\SS4U 4S jjjrS\R                  S\R                  4S jrS r	S	r
U =r$ )Mistral4RMSNorm1   epsreturnNc                    > [         TU ]  5         [        R                  " [        R
                  " U5      5      U l        X l        g)z.
Mistral4RMSNorm is equivalent to T5LayerNorm
N)super__init__r   	Parametertorchonesweightvariance_epsilon)selfhidden_sizer(   	__class__s      /root/GenerationalWealth/GenerationalWealth/venv/lib/python3.13/site-packages/transformers/models/mistral4/modeling_mistral4.pyr,   Mistral4RMSNorm.__init__3   s/     	ll5::k#:; #    hidden_statesc                    UR                   nUR                  [        R                  5      nUR	                  S5      R                  SSS9nU[        R                  " X0R                  -   5      -  nU R                  UR                  U5      -  $ )N   T)keepdim)	dtypetor.   float32powmeanrsqrtr1   r0   )r2   r8   input_dtypevariances       r5   forwardMistral4RMSNorm.forward;   sw    #))%((7 $$Q',,R,>%H?T?T4T(UU{{]--k:::r7   c                 ^    [        U R                  R                  5       SU R                   3$ )Nz, eps=)tupler0   shaper1   )r2   s    r5   
extra_reprMistral4RMSNorm.extra_reprB   s*    ))*+6$2G2G1HIIr7   )r1   r0   )gư>)__name__
__module____qualname____firstlineno__floatr,   r.   TensorrE   rJ   __static_attributes____classcell__r4   s   @r5   r&   r&   1   sB    $ $$ $ $;U\\ ;ell ;J Jr7   r&   c                      ^  \ rS rSr% \R
                  \S'   SS\4U 4S jjjr\	   SS\S-  S\
S   S\S-  S	\S
\4   4S jj5       r\R                  " 5       \S 5       5       rSrU =r$ )Mistral4RotaryEmbeddingF   inv_freqNconfigc                   > [         TU ]  5         UR                  U l        UR                  U l        Xl        U R
                  R                  S   U l        U R                  nU R                  S:w  a  [        U R                     nU" U R
                  U5      u  o@l
        U R                  SUSS9  U R                  SUR                  5       SS9  g )N	rope_typedefaultrX   F)
persistentoriginal_inv_freq)r+   r,   max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenrY   rope_parametersr[   compute_default_rope_parametersr   attention_scalingregister_bufferclone)r2   rY   devicerope_init_fnrX   r4   s        r5   r,    Mistral4RotaryEmbedding.__init__I   s    "("@"@$*$B$B!44[A!%!E!E>>Y&.t~~>L+7V+L((ZeD0(..2BuUr7   rg   ztorch.deviceseq_lenr)   ztorch.Tensorc           	         U R                   S   n[        U SS5      =(       d    U R                  U R                  -  nSnSU[        R
                  " SUS[        R                  S9R                  U[        R                  S9U-  -  -  nXe4$ )	aH  
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_thetahead_dimNg      ?r   r:   r=   )rg   r=   )	rb   getattrr3   num_attention_headsr.   arangeint64r>   rP   )rY   rg   rj   basedimattention_factorrX   s          r5   rc   7Mistral4RotaryEmbedding.compute_default_rope_parametersY   s    & %%l3fj$/c63E3EIcIc3c U\\!S!5;;?BB&X]XcXcBdgjjk
 ))r7   c                 L   U R                   S S S 2S 4   R                  5       R                  UR                  S   SS5      R	                  UR
                  5      nUS S 2S S S 24   R                  5       n[        UR
                  R                  [        5      (       a0  UR
                  R                  S:w  a  UR
                  R                  OSn[        USS9   UR                  5       UR                  5       -  R                  SS5      n[        R                  " Xf4SS	9nUR                  5       U R                  -  nUR                  5       U R                  -  n	S S S 5        WR	                  UR                   S
9W	R	                  UR                   S
94$ ! , (       d  f       N@= f)Nr   r;   r"   mpscpuF)device_typeenabledr:   rt   rn   )rX   rP   expandrI   r>   rg   
isinstancetypestrr   	transposer.   catcosrd   sinr=   )
r2   xposition_idsinv_freq_expandedposition_ids_expandedrz   freqsembr   r   s
             r5   rE   Mistral4RotaryEmbedding.forwardw   sN    !MM$4-8>>@GGHZHZ[\H]_acdehhijiqiqr ,QaZ 8 > > @'1!((--'E'E!((--[`J`ahhmmfkUC&,,.1F1L1L1NNYYZ[]^_E))UN3C'')d444C'')d444C	 D vvAGGv$cff177f&;;; DCs   BF
F#)rd   rY   r`   ra   r[   N)NNN)rL   rM   rN   rO   r.   rQ   __annotations__r#   r,   staticmethodr   intrH   rP   rc   no_gradr   rE   rR   rS   rT   s   @r5   rV   rV   F   s    llV~ V V  (,+/"*%*(* t* 
~u$	%	* *: ]]_<  <r7   rV   c                   2   ^  \ rS rSrSU 4S jjrS rSrU =r$ )Mistral4MLP   c                   > [         TU ]  5         Xl        UR                  U l        Uc  UR                  OUU l        [
        R                  " U R                  U R                  SS9U l        [
        R                  " U R                  U R                  SS9U l        [
        R                  " U R                  U R                  SS9U l	        [        UR                     U l        g NFbias)r+   r,   rY   r3   intermediate_sizer   Linear	gate_projup_proj	down_projr   
hidden_actact_fn)r2   rY   r   r4   s      r5   r,   Mistral4MLP.__init__   s    !--=N=V!9!9\m4#3#3T5K5KRWXyy!1!143I3IPUV4#9#94;K;KRWXV../r7   c                     U R                  U R                  U R                  U5      5      U R                  U5      -  5      nU$ r   )r   r   r   r   )r2   r   r   s      r5   rE   Mistral4MLP.forward   s6    NN4;;t~~a/@#ADLLQRO#ST	r7   )r   rY   r   r   r3   r   r   r   rL   rM   rN   rO   r,   rE   rR   rS   rT   s   @r5   r   r      s    0 r7   r   c                   .   ^  \ rS rSrU 4S jrS rSrU =r$ )Mistral4TopkRouter   c                    > [         TU ]  5         Xl        UR                  U l        [        R
                  " [        R                  " U R                  UR                  45      5      U l	        g r   )
r+   r,   rY   n_routed_expertsr   r-   r.   emptyr3   r0   r2   rY   r4   s     r5   r,   Mistral4TopkRouter.__init__   sK     & 7 7ll5;;0E0EvGYGY/Z#[\r7   c                     UR                  SU R                  R                  5      n[        R                  " XR
                  5      nU$ Nr;   )viewrY   r3   Flinearr0   )r2   r8   router_logitss      r5   rE   Mistral4TopkRouter.forward   s6    %**2t{{/F/FG<r7   )rY   r   r0   r   rT   s   @r5   r   r      s    ] r7   r   c                      ^  \ rS rSrSrU 4S jrS\R                  S\R                  S\R                  S\R                  4S jrS	r	U =r
$ )
Mistral4NaiveMoe   z2Collection of expert weights stored as 3D tensors.c                   > [         TU ]  5         UR                  U l        UR                  U l        UR                  U l        [        R                  " [        R                  " U R                  SU R                  -  U R
                  5      5      U l        [        R                  " [        R                  " U R                  U R
                  U R                  5      5      U l        [        UR                     U l        g )Nr:   )r+   r,   num_local_expertsnum_expertsr3   
hidden_dimmoe_intermediate_sizeintermediate_dimr   r-   r.   r   gate_up_projr   r   r   r   r   s     r5   r,   Mistral4NaiveMoe.__init__   s    !33 ,, & < <LLT5E5Eq4K`K`G`bfbqbq)rsekk$2B2BDOOUYUjUj&klV../r7   r8   top_k_indextop_k_weightsr)   c                 X   [         R                  " U5      n[         R                  " 5          [         R                  R                  R                  X R                  S9nUR                  SSS5      n[         R                  " UR                  SS9S5      R                  5       nS S S 5        W H  nUS   nXpR                  :X  a  M  [         R                  " WU   5      u  pX   n
[        R                  R                  XR                  U   5      R                  SSS9u  pU R                  U5      U-  n[        R                  R                  XR                   U   5      nXXS 4   -  nUR#                  SXR%                  UR&                  5      5        M     U$ ! , (       d  f       N= f)N)num_classesr:   r"   r   )r;   r|   r;   )r.   
zeros_liker   r   
functionalone_hotr   permutegreatersumnonzerowherer   r   chunkr   r   
index_add_r>   r=   )r2   r8   r   r   final_hidden_statesexpert_mask
expert_hit
expert_idx	top_k_pos	token_idxcurrent_stategateupcurrent_hidden_statess                 r5   rE   Mistral4NaiveMoe.forward   so    $..}=]]_((--55kO_O_5`K%--aA6K{8'DaHPPRJ 
 %J#AJ---#(;;{:/F#G I)4M}}++M;L;LZ;XY__`agi_jHD$(KK$5$:!$&MM$8$89NP^P^_iPj$k!$9)`dJd<e$e!**1i9Q9QReRkRk9lm % #"# _s   A7F
F))r   r   r   r   r   r   )rL   rM   rN   rO   __doc__r,   r.   rQ   rE   rR   rS   rT   s   @r5   r   r      sK    <0#||# \\# ||	#
 
# #r7   r   c                      ^  \ rS rSrSrU 4S jrS\R                  S\\R                  \R                  4   4S jr	S r
SrU =r$ )	Mistral4MoE   z2
A mixed expert module containing shared experts.
c                   > [         TU ]  5         Xl        [        U5      U l        [        U5      U l        [        XR                  UR                  -  S9U l
        UR                  U l        UR                  U l        UR                  U l        UR                  U l        UR                  U l        UR                   U l        g )N)rY   r   )r+   r,   rY   r   expertsr   r   r   r   n_shared_expertsshared_expertsr   n_group
topk_groupnorm_topk_probrouted_scaling_factornum_experts_per_toktop_kr   s     r5   r,   Mistral4MoE.__init__   s    '/&v.	)-I-IFLcLc-c
 !' 7 7~~ ++$33%+%A%A"//
r7   r   r)   c                 8   UR                  S5      nUR                  SU R                  U R                  U R                  -  5      R	                  SSS9S   R                  SS9n[        R                  " X R                  SSS9S   n[        R                  " U5      nUR                  SUS5        UR                  S5      R                  SU R                  U R                  U R                  -  5      R                  SU R                  5      nUR                  UR                  5       ) S5      n[        R                  " X`R                  SSS9S   nUR!                  SU5      nU R"                  (       a  UR                  SS	S
9S-   n	X-  nXR$                  -  nXx4$ )Nr;   r:   r|   r   F)krt   sortedr"           T)rt   r<   g#B;)softmaxr   r   r   topkr   r.   r   r   scatter_	unsqueezer}   reshapemasked_fillboolr   gatherr   r   )
r2   r   group_scores	group_idx
group_mask
score_maskscores_for_choicetopk_indicestopk_weightsdenominators
             r5   route_tokens_to_experts#Mistral4MoE.route_tokens_to_experts   sz   %--b1r4<<1F1F$,,1VW\\]^df\ghijnnsunv 	 JJ|BuUVWX	%%l3
Ay!,  $VBd&;&;t||&KLWR../ 	
 *55z7H6H#Nzz"3zzrRWXYZ[$++A|<&**r4*@5HK'L#&@&@@))r7   c                    UnUR                   nU R                  U5      nU R                  U5      u  pVUR                  SUR                   S   5      nU R	                  XU5      R                  " U6 nXR                  U5      -   nU$ r   )rI   r   r   r   r   r   )r2   r8   	residuals
orig_shaper   r   r   s          r5   rE   Mistral4MoE.forward   s    !	"((
		-0%)%A%A-%P"%**2}/B/B2/FG],OTTV`a%(;(;I(FFr7   )
rY   r   r   r   r   r   r   r   r   r   )rL   rM   rN   rO   r   r,   r.   rQ   rH   r   rE   rR   rS   rT   s   @r5   r   r      sC    0*U\\ *eELLZ_ZfZfLfFg *, r7   r   c                     U SSU R                   S   S-  24   nU SU R                   S   S-  S24   n[        R                  " U* U4SS9$ )z*Rotates half the hidden dims of the input..Nr;   r:   r|   )rI   r.   r   )r   x1x2s      r5   rotate_halfr    sZ    	
3"!''"+"""	#B	
3q ""	#B99rc2YB''r7   rotary_pos_embc                     UR                  U5      nUR                  U5      nX-  [        U 5      U-  -   nX-  [        U5      U-  -   nXV4$ )aI  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.
)r   r  )qr   r   r   unsqueeze_dimq_embedk_embeds          r5   apply_rotary_pos_embr
  	  sS    & --
&C
--
&Cw;q>C/0Gw;q>C/0Gr7   r8   n_repr)   c                     U R                   u  p#pEUS:X  a  U $ U SS2SS2SSS2SS24   R                  X#XU5      n U R                  X#U-  XE5      $ )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)rI   r}   r   )r8   r  batchnum_key_value_headsslenrm   s         r5   	repeat_kvr  #  s_    
 2?1D1D.Ez!!Qa"23::5W\dlmM  e(CTTTr7   modulequerykeyvalueattention_maskscalingdropoutkwargsc                    [        X R                  5      n[        X0R                  5      n	[        R                  " XR	                  SS5      5      U-  n
Ub  X-   n
[
        R                  R                  U
S[        R                  S9R                  UR                  5      n
[
        R                  R                  XU R                  S9n
[        R                  " X5      nUR	                  SS5      R                  5       nX4$ )Nr:   r   r;   )rt   r=   )ptrainingr"   )r  num_key_value_groupsr.   matmulr   r   r   r   r?   r>   r=   r  r  
contiguous)r  r  r  r  r  r  r  r  
key_statesvalue_statesattn_weightsattn_outputs               r5   eager_attention_forwardr#  /  s     3 ; ;<JU$?$?@L<<';';Aq'ABWLL!#4==((2U]](SVVW\WbWbcL==((6??([L,,|:K''1-88:K$$r7   c                    UR                  U5      nUR                  U5      nU R                  u  pgpU R                  XgXS-  S5      R                  SS5      R	                  XgX5      n UR                  u  pgpUR                  XgXS-  S5      R                  SS5      R	                  XgX5      nX-  [        U 5      U-  -   n
X-  [        U5      U-  -   nX4$ )a  
TODO let's just use the original freqcis computation to not have the view
transpose + reshape! This is not optimized!
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.
    position_ids (`torch.Tensor`):
        The position indices of the tokens corresponding to the query and key tensors. For example, this can be
        used to pass offsetted position ids when working with a KV-cache.
    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.
r:      r   )r   rI   r   r   r   r  )r  r   r   r   r   r  bhsdr  r	  s               r5   apply_rotary_pos_emb_interleaver*  H  s    0 --
&C
--
&CJA!	qQQ",,Q2::1FAJA!	qQQ",,Q2::1FAw;q>C/0Gw;q>C/0Gr7   positions_idsbetar_   c           	          SU[         R                  " S[         R                  " X-  5      -   5      -  -   nUS S 2S S S 2S 4   $ Nr"   )r.   logfloor)r+  r,  r_   r  s       r5   get_llama_4_attn_scaler1  n  s?    $1u{{=3Z'[#[\\\G1dAt#$$r7   c                   >  ^  \ rS rSrSrS\S\4U 4S jjr SS\R                  S\
\R                  \R                  4   S	\R                  S-  S
\R                  S\S-  S\\   S\
\R                  \R                  S-  \
\R                     S-  4   4S jjrSrU =r$ )Mistral4Attentionis  z=Multi-headed attention from 'Attention Is All You Need' paperrY   	layer_idxc                 L  > [         TU ]  5         Xl        X l        UR                  UR
                  -  U l        UR                  U l        UR                  U l        UR                  U l	        UR                  U l
        UR                  U l        UR                  U l        UR                  U l        UR                  U l        SU l        U R                  c=  [         R"                  " UR$                  U R                  U R                  -  SS9U l        O[         R"                  " UR$                  UR                  UR(                  S9U l        [-        UR                  5      U l        [         R"                  " UR                  U R                  U R                  -  SS9U l        [         R"                  " UR$                  U R                  U R                  -   UR(                  S9U l        [-        U R                  5      U l        [         R"                  " U R                  U R                  U R                  U R                  -   -  SS9U l        [         R"                  " U R                  U R                  -  UR$                  UR(                  S9U l        U R                  S-  U l        g )NTFr   g      )r+   r,   rY   r4  rp   r  r  attention_dropout	num_headsq_lora_rankqk_rope_head_dimkv_lora_rank
v_head_dimqk_nope_head_dimqk_head_dim	is_causalr   r   r3   q_projattention_biasq_a_projr&   q_a_layernormq_b_projkv_a_proj_with_mqakv_a_layernorm	kv_b_projo_projr  r2   rY   r4  r4   s      r5   r,   Mistral4Attention.__init__v  s   "$*$>$>&B\B\$\!!'!9!933!-- & 7 7"// ++ & 7 7!--#))F$6$6IYIY8Y`efDKIIf&8&8&:L:LSYShShiDM!01C1C!DDIIf&8&8$..4K[K[:[bghDM"$)) 5 55&&#

 .d.?.?@NNd33dooEF
 iiNNT__,&&
 ''D1r7   Nr8   position_embeddingsr  r   past_key_valuesr  r)   c                 ,   UR                   S S u  pxXxSU R                  4n	XxSU R                  U R                  -   4n
U R                  c  U R                  U5      nO/U R                  U R                  U R                  U5      5      5      nUR                  U	5      R                  SS5      n[        R                  " XR                  U R                  /SS9u  pU R                  U5      n[        R                  " XR                  U R                  /SS9u  nnU R!                  U R#                  U5      5      R                  U
5      R                  SS5      n[        R                  " XR                  U R                  /SS9u  nnUR                  USXR                  5      nUu  nnU R$                  R&                  (       a  [)        UUUU5      u  nnO[+        UUUU5      u  nnUR,                  " / UR                   S S QSP76 n[        R.                  " X4SS9n[        R.                  " UU4SS9nU[1        UU R$                  R2                  R5                  S5      U R$                  R2                  R5                  S5      5      R7                  UR8                  5      -  nUb   UR;                  UUU R<                  5      u  nn[?        U R$                  5      (       aJ  U R                  U R                  :w  a0  [@        RB                  " USU R                  U R                  -
  /5      n[D        RF                  " U R$                  RH                  [J        5      nU" U UUUU4U RL                  (       d  SOU RN                  U RP                  S	.UD6u  nn[?        U R$                  5      (       a5  U R                  U R                  :w  a  US S 2S S 2S S 2S U R                  24   nURS                  XxS5      RU                  5       nU RW                  U5      nUU4$ )
Nr;   r"   r:   r|   llama_4_scaling_beta original_max_position_embeddingsr   r   )r  r  ),rI   r=  r<  r;  r8  r?  rC  rB  rA  r   r   r.   splitr9  rD  r:  rF  rE  rY   rope_interleaver*  r
  r}   r   r1  rb   getr>   r=   updater4  r   r   padr   get_interface_attn_implementationr#  r  r6  r  r   r  rG  )r2   r8   rJ  r  r   rK  r  
batch_size
seq_lengthquery_shape	key_shapeq_statesq_passq_rotcompressed_kvk_passk_rotr   r   r   query_statesr  attention_interfacer"  r!  s                            r5   rE   Mistral4Attention.forward  s    "/!4!4Sb!9
!r43C3CDR1F1F1XY	#{{=1H}}T%7%7m8T%UVH==-771=H/D/DdF[F[.\bde//>M4E4EtG\G\3]cef 3 3F ;<AA)LVVWXZ[\${{64I4I4??3[acd

:q*6K6KL&S;;&&:5%cRLE5/uc3GLE54fll3B/44yy&b9YYB7
#&<KK''++,BCKK''++,NO'
 "\
 	! &'6'='=j,X\XfXf'g$J'449I9IT__9\5543C3Cdoo3U/VWL(?(M(MKK,,.E)
 %8	%
  $}}C$2H2HLL	%
 	%
!\ (449I9IT__9\%aA/@/@&@AK!))*"EPPRkk+.L((r7   )r6  rY   r>  rE  rD  rF  r:  r4  r7  r  rG  rB  rA  rC  r8  r?  r=  r<  r9  r  r;  r   )rL   rM   rN   rO   r   r#   r   r,   r.   rQ   rH   r	   r   r   rE   rR   rS   rT   s   @r5   r3  r3  s  s    G)2~ )2# )2b )-F)||F) #5<<#=>F) t+	F)
 llF) F) -.F) 
u||U\\D0%2E2LL	MF) F)r7   r3  c                     ^  \ rS rSrS\S\4U 4S jjr     SS\R                  S\R                  S-  S\R                  S-  S	\
S-  S
\S-  S\\R                  \R                  4   S-  S\\   S\R                  4S jjrSrU =r$ )Mistral4DecoderLayeri  rY   r4  c                 L  > [         TU ]  5         UR                  U l        [        XS9U l        X!R
                  :  a  [        U5      U l        O[        U5      U l        [        UR                  UR                  S9U l        [        UR                  UR                  S9U l        g )N)rY   r4  r(   )r+   r,   r3   r3  	self_attnfirst_k_dense_replacer   mlpr   r&   rms_norm_epsinput_layernormpost_attention_layernormrH  s      r5   r,   Mistral4DecoderLayer.__init__  s    !--*&N444"6*DH"6*DH.v/A/AvGZGZ[(78J8JPVPcPc(d%r7   Nr8   r  r   rK  	use_cacherJ  r  r)   c           
          UnU R                  U5      nU R                  " SUUUUUUS.UD6u  pX-   nUnU R                  U5      nU R                  U5      nX-   nU$ )N)r8   r  r   rK  rn  rJ   )rk  rg  rl  ri  )
r2   r8   r  r   rK  rn  rJ  r  residual_s
             r5   rE   Mistral4DecoderLayer.forward  s     !,,];>> 
')%+ 3
 
 !0 !55mD/ 0r7   )r3   rk  ri  rl  rg  )NNNFN)rL   rM   rN   rO   r#   r   r,   r.   rQ   
LongTensorr	   r   rH   r   r   rE   rR   rS   rT   s   @r5   rd  rd    s    e~ e# e" /304(,!&HL|| t+ &&-	
  $; #5<<#=>E +, 
 r7   rd  c                      ^  \ rS rSr% \\S'   SrSrS/rS/r	Sr
SrSrSrSr\\S.r/ r/ r\R*                  " 5       U 4S j5       rS	rU =r$ )
Mistral4PreTrainedModeli  rY   modelTrd  rK  )r8   
attentionsc                   > [         TU ]  U5        [        U[        5      (       a5  [        R
                  " UR                  SU R                  R                  S9  g [        U[        5      (       ai  [        R
                  " UR                  SU R                  R                  S9  [        R
                  " UR                  SU R                  R                  S9  g g )Nr   )rA   std)r+   _init_weightsr~   r   initnormal_r0   rY   initializer_ranger   r   r   )r2   r  r4   s     r5   r{  %Mistral4PreTrainedModel._init_weights,  s    f%f011LLSdkk6S6ST 011LL,,3DKK<Y<YZLL))9V9VW 2r7   rp  )rL   rM   rN   rO   r#   r   base_model_prefixsupports_gradient_checkpointing_no_split_modules_skip_keys_device_placement_supports_flash_attn_supports_sdpa_supports_flex_attn_can_compile_fullgraph_supports_attention_backendrd  r3  _can_record_outputs_keep_in_fp32_modules_strict"_keys_to_ignore_on_load_unexpectedr.   r   r{  rR   rS   rT   s   @r5   rv  rv    sz    &*#/0#4"5N!"&-' $& )+&
]]_X Xr7   rv  c                     ^  \ rS rSrS\4U 4S jjr\\\      SS\	R                  S-  S\	R                  S-  S\	R                  S-  S\S-  S	\	R                  S-  S
\S-  S\\   S\4S jj5       5       5       rSrU =r$ )Mistral4Modeli6  rY   c           	        > [         TU ]  U5        UR                  U l        UR                  U l        [
        R                  " UR                  UR                  U R                  5      U l        [
        R                  " [        UR                  5       Vs/ s H  n[        X5      PM     sn5      U l        [        UR                  UR                  S9U l        [#        US9U l        SU l        U R)                  5         g s  snf )Nrf  rY   F)r+   r,   pad_token_idpadding_idx
vocab_sizer   	Embeddingr3   embed_tokens
ModuleListrangenum_hidden_layersrd  layersr&   rj  normrV   
rotary_embgradient_checkpointing	post_initrH  s      r5   r,   Mistral4Model.__init__8  s     !.. ++LL):):F<N<NPTP`P`ammFKFLdLdFefFe!&4Fef
 $F$6$6F<O<OP	1@&+# 	 gs   C?N	input_idsr  r   rK  inputs_embedsrn  r  r)   c           
      >   US L US L-  (       a  [        S5      eUc  U R                  U5      nU(       a  Uc  [        U R                  S9nUcU  Ub  UR	                  5       OSn[
        R                  " UR                  S   UR                  S9U-   nUR                  S5      n[        U R                  UUUUS9n	Un
U R                  XS9nU R                  S U R                  R                    H  nU" U
4U	UUUUS.UD6n
M     U R                  U
5      n
[        U
US	9$ )
Nz:You must specify exactly one of input_ids or inputs_embedsr  r   r"   )rg   )rY   r  r  rK  r   )r   )r  rJ  r   rK  rn  )last_hidden_staterK  )
ValueErrorr  r
   rY   get_seq_lengthr.   rq   rI   rg   r   r   r  r  r  r  r   )r2   r  r  r   rK  r  rn  r  past_seen_tokenscausal_maskr8   rJ  decoder_layers                r5   rE   Mistral4Model.forwardH  sF    -t";<YZZ *.*;*;I*FM0*$++>OCRC^==?de <<(;(;A(>}G[G[\_ooL'11!4L(;;')+%
 &"oomoW![[)H4;;+H+HIM)*$7) /# M J 		-0&++
 	
r7   )r  r  r  r  r  r  r  )NNNNNN)rL   rM   rN   rO   r#   r,   r    r!   r   r.   rt  rQ   r	   FloatTensorr   r   r   r   rE   rR   rS   rT   s   @r5   r  r  6  s    ~     .2.204(,26!%2
##d*2
 t+2
 &&-	2

 2
 ((4/2
 $;2
 +,2
 
!2
    2
r7   r  c                   P  ^  \ rS rSrSS0rSS0rSS/S/40rU 4S jr\\	        SS
\
R                  S	-  S\
R                  S	-  S\
R                  S	-  S\S	-  S\
R                  S	-  S\
R                  S	-  S\S	-  S\\
R                  -  S\\   S\4S jj5       5       rSrU =r$ )Mistral4ForCausalLMi  zlm_head.weightzmodel.embed_tokens.weightlm_headcolwise_gather_outputr8   logitsc                    > [         TU ]  U5        [        U5      U l        UR                  U l        [
        R                  " UR                  UR                  SS9U l        U R                  5         g r   )
r+   r,   r  rw  r  r   r   r3   r  r  r   s     r5   r,   Mistral4ForCausalLM.__init__  sU     "6*
 ++yy!3!3V5F5FUS 	r7   Nr  r  r   rK  r  labelsrn  logits_to_keepr  r)   c	           
      |   U R                   " SUUUUUUS.U	D6n
U
R                  n[        U[        5      (       a  [	        U* S5      OUnU R                  USS2USS24   5      nSnUb)  U R                  " SXU R                  R                  S.U	D6n[        UUU
R                  U
R                  U
R                  S9$ )a  
Example:

```python
>>> from transformers import AutoTokenizer, Mistral4ForCausalLM

>>> model = Mistral4ForCausalLM.from_pretrained("meta-mistral4/Mistral4-2-7b-hf")
>>> tokenizer = AutoTokenizer.from_pretrained("meta-mistral4/Mistral4-2-7b-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  r  r   rK  r  rn  N)r  r  r  )lossr  rK  r8   rx  rp  )rw  r  r~   r   slicer  loss_functionrY   r  r   rK  r8   rx  )r2   r  r  r   rK  r  r  rn  r  r  outputsr8   slice_indicesr  r  s                  r5   rE   Mistral4ForCausalLM.forward  s    > ,0:: ,
)%+',
 ,
  118B>SV8W8W~ot4]kmA}a,?@A%%pVt{{OeOepiopD%#33!//))
 	
r7   )r  rw  r  )NNNNNNNr   )rL   rM   rN   rO   _tied_weights_keys_tp_plan_pp_planr,   r   r   r.   rt  rQ   r	   r  r   r   r   r   r   rE   rR   rS   rT   s   @r5   r  r    s   *,GH23H_-z:;H  .2.204(,26*.!%-.6
##d*6
 t+6
 &&-	6

 6
 ((4/6
   4'6
 $;6
 ell*6
 +,6
 
 6
  6
r7   r  c                       \ rS rSrSrg)!Mistral4ForSequenceClassificationi  rp  NrL   rM   rN   rO   rR   rp  r7   r5   r  r        r7   r  c                       \ rS rSrSrg)Mistral4ForTokenClassificationi  rp  Nr  rp  r7   r5   r  r    r  r7   r  )rv  r  r  r  r  )r"   )r   r.  )Ncollections.abcr   typingr   r.   torch.nn.functionalr   r   r    r   r|  activationsr   cache_utilsr	   r
   
generationr   integrationsr   r   r   masking_utilsr   modeling_flash_attention_utilsr   modeling_layersr   r   r   modeling_outputsr   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   utils.genericr   r   r    utils.output_capturingr!   configuration_mistral4r#   Moduler&   rV   r   r   r   r   r  r
  rQ   r   r  rP   r#  r*  r1  r3  rd  rv  r  r  r  r  __all__rp  r7   r5   <module>r     sj  ( %      & ! . ) m m / B 
 P K F & I I e e 5 2 Y'Jbii J (J(><bii ><B"))    $#ryy $# $#N2")) 2j( *+ ,2	UU\\ 	U# 	U%,, 	U& %II%<<% 
% <<	%
 LL4'% % % '(%2#L%%,, %e %^a %fkfrfr %
t)		 t)n,5 ,^Xo X: F
+ F
 F
R F
1? F
 F
R	(HJa 		%BD[ 	r7   