
    Z j                        S SK Jr  S SKJrJr  S SKrS SK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  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)  SSK*J+r+J,r,J-r-  SSK.J/r/  SSK0J1r1   " S S\SS9r2 " S S\Rf                  5      r4\" S5       " S S\Rf                  5      5       r5 " S  S!\Rf                  5      r6 " S" S#\Rf                  5      r7 " S$ S%\Rf                  5      r8S& r9\" S'5      SFS( j5       r:S)\Rv                  S*\<S+\Rv                  4S, jr= SGS-\Rf                  S.\Rv                  S/\Rv                  S0\Rv                  S1\Rv                  S-  S2\>S3\>S4\&\(   4S5 jjr?\" \:5       " S6 S7\Rf                  5      5       r@ " S8 S9\5      rA\) " S: S;\$5      5       rB " S< S=\Rf                  5      rC\) " S> S?\B5      5       rD   SHS@\Rv                  \E\Rv                     -  S-  SA\<S-  S1\Rv                  S-  S+\Rv                  \<-  4SB jjrF\) " SC SD\B\5      5       rG/ SEQrHg)I    )Callable)Optional	TypedDictN)nn)
functional   )initialization)ACT2FN)CacheDynamicCache)GenerationMixin)use_kernel_forward_from_hubuse_kernel_func_from_hubuse_kernelized_func)create_causal_mask)GradientCheckpointingLayer)MoeCausalLMOutputWithPastMoeModelOutputWithPast)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   )GraniteMoeSharedConfigc                       \ rS rSr% Sr\R                  \S'   \R                  \S'   \\S'   \\S'   \R                  \S'   Sr
g	)
GraniteFlashAttentionKwargs-   a   
Keyword arguments for advanced Flash Attention, causal-conv1d, and mamba_ssm kernel usage.
Use cases include padding-free training and fewer `torch.compile` graph breaks.

cu_seq_lens_q (`torch.LongTensor`):
    Gets cumulative sequence length for query state.
cu_seq_lens_k (`torch.LongTensor`):
    Gets cumulative sequence length for key state.
max_length_q (`int`):
    Maximum sequence length for query state.
max_length_k (`int`):
    Maximum sequence length for key state.
seq_idx (`torch.IntTensor):
    Index of each packed sequence.
cu_seq_lens_qcu_seq_lens_kmax_length_qmax_length_kseq_idx N)__name__
__module____qualname____firstlineno____doc__torch
LongTensor__annotations__int	IntTensor__static_attributes__r*       ڏ/root/GenerationalWealth/GenerationalWealth/venv/lib/python3.13/site-packages/transformers/models/granitemoeshared/modeling_granitemoeshared.pyr#   r#   -   s7      ######__r6   r#   F)totalc                   n   ^  \ rS rSrSrS\4U 4S jjrS\R                  S\R                  4S jr	Sr
U =r$ )	GraniteMoeSharedMLPE   zj
MLP layer for shared experts

Args:
    config:
        Configuration object with model hyperparameters.
configc                 X  > [         TU ]  5         UR                  U l        UR                  U l        [
        UR                     U l        [        R                  " U R                  U R                  S-  SS9U l
        [        R                  " U R                  U R                  SS9U l        g )N   Fbias)super__init__hidden_size
input_sizeshared_intermediate_sizer
   
hidden_act
activationr   Linearinput_linearoutput_linearselfr<   	__class__s     r7   rB   GraniteMoeSharedMLP.__init__N   s     ,,!:: !2!23IIdoot7G7G!7KRWXYYt'7'7uUr6   hidden_statesreturnc                     U R                  U5      nUR                  SSS9nU R                  US   5      US   -  nU R                  U5      nU$ )Nr>   dimr   r    )rI   chunkrG   rJ   )rL   rO   chunked_hidden_statess      r7   forwardGraniteMoeSharedMLP.forwardW   s^    ))-8 - 3 3A2 3 >(=a(@ADYZ[D\\**=9r6   )rG   rC   rI   rD   rJ   )r+   r,   r-   r.   r/   r!   rB   r0   TensorrW   r5   __classcell__rM   s   @r7   r:   r:   E   s7    V5 VU\\ ell  r6   r:   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$ )GraniteMoeSharedRMSNorm_   epsrP   Nc                    > [         TU ]  5         [        R                  " [        R
                  " U5      5      U l        X l        g)z6
GraniteMoeSharedRMSNorm is equivalent to T5LayerNorm
N)rA   rB   r   	Parameterr0   onesweightvariance_epsilon)rL   rC   r`   rM   s      r7   rB    GraniteMoeSharedRMSNorm.__init__a   s/     	ll5::k#:; #r6   rO   c                    UR                   nUR                  [        R                  5      nUR	                  S5      R                  SSS9nU[        R                  " X0R                  -   5      -  nU R                  UR                  U5      -  $ )Nr>   rR   T)keepdim)	dtypetor0   float32powmeanrsqrtre   rd   )rL   rO   input_dtypevariances       r7   rW   GraniteMoeSharedRMSNorm.forwardi   sw    #))%((7 $$Q',,R,>%H?T?T4T(UU{{]--k:::r6   c                 ^    [        U R                  R                  5       SU R                   3$ )Nz, eps=)tuplerd   shapere   )rL   s    r7   
extra_repr"GraniteMoeSharedRMSNorm.extra_reprp   s*    ))*+6$2G2G1HIIr6   )re   rd   )gư>)r+   r,   r-   r.   floatrB   r0   rY   rW   ru   r5   rZ   r[   s   @r7   r^   r^   _   sB    $ $$ $ $;U\\ ;ell ;J Jr6   r^   c                   B   ^  \ rS rSrS\S\S\SS4U 4S jjrS rS	rU =r$ )
GraniteMoeSharedParallelExpertst   num_expertsrD   output_sizerP   Nc                    > [         TU ]  5         [        R                  " [        R
                  " XU5      5      U l        Xl        X l        X0l	        g)a]  
Initialize the GraniteMoeSharedParallelExperts module.
The experts weights are stored in [num_experts, output_size, input_size] format. Such that it's compatible with
many MoE libraries, such as [Megablock](https://github.com/databricks/megablocks) and
[ScatterMoE](https://github.com/shawntan/scattermoe), as well as the
[MoE kernel](https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/layers/fused_moe/fused_moe.py)
used in vllm.

Args:
    num_experts (int):
        Number of experts.
    input_size (int):
        Size of the input.
    output_size (int):
        Size of the output.
N)
rA   rB   r   rb   r0   emptyrd   r{   rD   r|   )rL   r{   rD   r|   rM   s       r7   rB   (GraniteMoeSharedParallelExperts.__init__u   s<    " 	ll5;;{#TU&$&r6   c                     UR                  USS9n/ n[        U R                  5       H8  nUR                  [        R
                  " X5   U R                  U   5      5        M:     [        R                  " USS9nU$ )z
Forward pass of the GraniteMoeSharedParallelExperts module.

Args:
    inputs (Tensor):
        Input tensor.
    expert_size:
        Expert size information.

Returns:
    Tensor: Output tensor.
r   rS   )	splitranger{   appendFlinearrd   r0   cat)rL   inputsexpert_size
input_listoutput_listiresultss          r7   rW   'GraniteMoeSharedParallelExperts.forward   sh     \\+1\5
t''(Aqxx
t{{1~FG )))KQ/r6   )rD   r{   r|   rd   	r+   r,   r-   r.   r3   rB   rW   r5   rZ   r[   s   @r7   ry   ry   t   s.    'C 'S 's 't '. r6   ry   c                   >   ^  \ rS rSrS\S\S\4U 4S jjrS rSrU =r$ )GraniteMoeSharedTopKGating   rD   r{   top_kc                 z   > [         TU ]  5         X l        Xl        X0l        [
        R                  " XSS9U l        g)z
Initialize the top-k gating mechanism.

Args:
    input_size (`int`):
        Size of the input.
    num_experts (`int`):
        Number of experts.
    top_k (`int`):
        Number of top experts to select.
Fr?   N)rA   rB   r{   rD   r   r   rH   layer)rL   rD   r{   r   rM   s       r7   rB   #GraniteMoeSharedTopKGating.__init__   s2     	&$
YYzUC
r6   c                 z   U R                  U5      R                  5       nUR                  U R                  SS9u  p4[        R
                  " USS9R                  U5      n[        R                  " UR                  S5      U R                  /UR                  UR                  S9nUR                  SUS5      nUR                  5       R                  S5      nUR                  5       nUR!                  5       n	U	R#                  S5      u  pUR%                  U R                  SS9nUR!                  5       nX[   nXXU4$ )Nr    rS   r   ri   devicetrunc)rounding_mode)r   rw   topkr   r0   softmaxtype_aszerossizer{   ri   r   scatterlongsumtolistflattensortdiv)rL   rO   logitstop_k_logitstop_k_indicestop_k_gatesr   gatesr   top_k_experts_index_sorted_expertsbatch_indexbatch_gatess                 r7   rW   "GraniteMoeSharedTopKGating.forward   s"   M*002&,kk$**!k&D#mmLa8@@O a $"2"23;;L;LU`UgUg
 a2jjl&&q) "((* &--/"/"4"4Q"7*..tzz.Q "))+!7#+FRRr6   )rD   r   r{   r   r   r[   s   @r7   r   r      s-    D3 DS D D(S Sr6   r   c                   :   ^  \ rS rSrSrS\4U 4S jjrS rSrU =r	$ )GraniteMoeSharedMoE   z
A Sparsely gated mixture of experts layer with 1-layer Feed-Forward networks as experts.

Args:
    config:
        Configuration object with model hyperparameters.
r<   c                   > [         TU ]  5         UR                  U l        UR                  U l        [
        UR                     U l        [        UR                  U R                  U R                  S-  5      U l
        [        UR                  U R                  U R                  5      U l        [        U R                  UR                  UR                  S9U l        g )Nr>   )rD   r{   r   )rA   rB   rC   rD   intermediate_sizer
   rF   rG   ry   num_local_expertsrI   rJ   r   num_experts_per_tokrouterrK   s     r7   rB   GraniteMoeSharedMoE.__init__   s     ,,!33 !2!23;$$doot7G7G!7K
 =$$d&6&6
 100,,
r6   c                    UR                  5       u  p#nUR                  SU5      nU R                  U5      u  pVpxnX   n	U R                  X5      n
U
R	                  SSS9nU R                  US   5      US   -  n
U R                  X5      nXS S 2S 4   -  n[        R                  " X#-  U R                  4UR                  UR                  S9nUR                  SXl5      nUR                  X#U R                  5      nU$ )NrR   r>   rS   r   r    r   )r   reshaper   rI   rU   rG   rJ   r0   r   rD   ri   r   	index_addview)rL   layer_inputbszlengthemb_sizer   r   r   r   expert_inputsrO   rV   expert_outputsr   layer_outputs                  r7   rW   GraniteMoeSharedMoE.forward   s    + 0 0 2X!))"h76:kk+6N3!#0))-E - 3 3A2 3 >(=a(@ADYZ[D\\++MG'ag*>>S\4??;>CWCW`n`u`uvq+F#((dooFr6   )rG   rC   rI   rD   rJ   r   )
r+   r,   r-   r.   r/   r!   rB   rW   r5   rZ   r[   s   @r7   r   r      s    
5 
& r6   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$ )z*Rotates half the hidden dims of the input..NrR   r>   rS   )rt   r0   r   )xx1x2s      r7   rotate_halfr     sZ    	
3"!''"+"""	#B	
3q ""	#B99rc2YB''r6   rotary_pos_embc                     UR                  U5      nUR                  U5      nX-  [        U 5      U-  -   nX-  [        U5      U-  -   nXV4$ )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.
)	unsqueezer   )qkcossinunsqueeze_dimq_embedk_embeds          r7   apply_rotary_pos_embr     sS    & --
&C
--
&Cw;q>C/0Gw;q>C/0Gr6   rO   n_reprP   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)rt   expandr   )rO   r   batchnum_key_value_headsslenhead_dims         r7   	repeat_kvr   "  s_    
 2?1D1D.Ez!!Qa"23::5W\dlmM  e(CTTTr6   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$ )Nr>   r   rR   )rT   ri   )ptrainingr    )r   num_key_value_groupsr0   matmul	transposer   r   r   rk   rj   ri   r   r   
contiguous)r   r   r   r   r   r   r   r   
key_statesvalue_statesattn_weightsattn_outputs               r7   eager_attention_forwardr   .  s     3 ; ;<JU$?$?@L<<';';Aq'ABWLL!#4==((2U]](SVVW\WbWbcL==((6??([L,,|:K''1-88:K$$r6   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-  S	\R                  S-  S
\S-  S\\   S\
\R                  \R                  4   4S jjrSrU =r$ )GraniteMoeSharedAttentioniG  z=Multi-headed attention from 'Attention Is All You Need' paperr<   	layer_idxc                 J  > [         TU ]  5         Xl        X l        [	        USUR
                  UR                  -  5      U l        UR                  UR                  -  U l	        UR                  U l        UR                  U l        SU l        [        R                  " UR
                  UR                  U R                  -  UR                   S9U l        [        R                  " UR
                  UR                  U R                  -  UR                   S9U l        [        R                  " UR
                  UR                  U R                  -  UR                   S9U l        [        R                  " UR                  U R                  -  UR
                  UR                   S9U l        g )Nr   Tr?   )rA   rB   r<   r   getattrrC   num_attention_headsr   r   r   attention_multiplierr   attention_dropout	is_causalr   rH   attention_biasq_projk_projv_projo_projrL   r<   r   rM   s      r7   rB   "GraniteMoeSharedAttention.__init__K  sF   "
F4F4F&JdJd4de$*$>$>&B\B\$\!22!'!9!9ii : :T]] JQWQfQf
 ii : :T]] JQWQfQf
 ii : :T]] JQWQfQf
 ii&&68J8JQWQfQf
r6   NrO   position_embeddingsr   past_key_valuesr   rP   c                    UR                   S S n/ UQSPU R                  P7nU R                  U5      R                  U5      R	                  SS5      nU R                  U5      R                  U5      R	                  SS5      n	U R                  U5      R                  U5      R	                  SS5      n
Uu  p[        XX5      u  pUb  UR                  XU R                  5      u  p[        R                  " U R                  R                  [        5      nU" U UU	U
U4U R                  (       d  SOU R                   U R"                  S.UD6u  pUR$                  " / UQSP76 R'                  5       nU R)                  U5      nX4$ )NrR   r    r>           )r   r   )rt   r   r   r   r   r   r   r   updater   r   get_interfacer<   _attn_implementationr   r   r   r   r   r   r   )rL   rO   r  r   r  r   input_shapehidden_shapequery_statesr   r   r   r   attention_interfacer   r   s                   r7   rW   !GraniteMoeSharedAttention.forwardb  s~    $))#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&#7RU#[ &'6'='=jX\XfXf'g$J(?(M(MKK,,.E)
 %8	%
  $}}C$2H2HLL	%
 	%
! "));;;;FFHkk+.((r6   )r   r<   r   r   r   r   r   r   r   r   r   NNN)r+   r,   r-   r.   r/   r!   r3   rB   r0   rY   rs   r   r   r   rW   r5   rZ   r[   s   @r7   r   r   G  s    G
5 
# 
4 IM.2(,&)||&) #5<<#=>E&) t+	&)
 &) +,&) 
u||U\\)	*&) &)r6   r   c                   `  ^  \ rS rSrS\S\4U 4S jjr      SS\R                  S\R                  S-  S\R                  S-  S	\
S-  S
\S-  S\S-  S\\R                  \R                  4   S-  S\\   S\\R                  \\R                  \R                  4   S-  4   4S jjrSrU =r$ )GraniteMoeSharedDecoderLayeri  r<   r   c                 ~  > [         TU ]  5         UR                  U l        [        XS9U l        [        UR                  UR                  S9U l        [        UR                  UR                  S9U l        [        U5      U l
        UR                  U l        UR                  S:X  a  S U l        g [        U5      U l        g )N)r<   r   r`   r   )rA   rB   rC   r   	self_attnr^   rms_norm_epsinput_layernormpost_attention_layernormr   block_sparse_moeresidual_multiplierrE   r:   
shared_mlpr  s      r7   rB   %GraniteMoeSharedDecoderLayer.__init__  s    !--2&V6v7I7IvObObc(?@R@RX^XkXk(l% 3F ;#)#=#= "("A"AQ"F$L_`fLgr6   NrO   r   position_idsr  output_attentions	use_cacher  r   rP   c                 6   Un	U R                  U5      nU R                  " SUUUUUUUS.UD6u  pXU R                  -  -   nUn	U R                  U5      nU R	                  U5      nU R
                  c  UnOXR                  U5      -   nXU R                  -  -   nU$ )N)rO   r   r  r  r  r  r  r*   )r  r  r  r  r  r  )rL   rO   r   r  r  r  r  r  r   residualr   moe_hidden_statess               r7   rW   $GraniteMoeSharedDecoderLayer.forward  s     !,,];  >> 	
')%+/ 3	
 	
 !43K3K#KK 55mD 11-@??"-M-0NNM 43K3K#KKr6   )r  rC   r  r  r  r  r  )NNNFFN)r+   r,   r-   r.   r!   r3   rB   r0   rY   r1   r   boolrs   r   r#   FloatTensorrW   r5   rZ   r[   s   @r7   r  r    s    h5 h# h /304(,).!&HL%||% t+% &&-	%
 %  $;% $;% #5<<#=>E% 45% 
u  %(9(95;L;L(L"MPT"TT	U% %r6   r  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&                  " 5       U 4S	 j5       rS
rU =r$ )GraniteMoeSharedPreTrainedModeli  r<   modelTr  r  F)rO   
attentionsc                    > [         TU ]  U5        [        U[        5      (       a5  [        R
                  " UR                  SU R                  R                  S9  g g )Nr  )rm   std)	rA   _init_weights
isinstancery   initnormal_rd   r<   initializer_range)rL   r   rM   s     r7   r+  -GraniteMoeSharedPreTrainedModel._init_weights  sA    f%f=>>LLSdkk6S6ST ?r6   r*   )r+   r,   r-   r.   r!   r2   base_model_prefixsupports_gradient_checkpointing_no_split_modules_skip_keys_device_placement_supports_flash_attn_supports_sdpa_supports_flex_attn_can_compile_fullgraph_supports_attention_backendr  r   _can_record_outputsr0   no_gradr+  r5   rZ   r[   s   @r7   r&  r&    sn    ""&*#78#4"5N""&5/
 ]]_U Ur6   r&  c                      ^  \ rS rSr% \R
                  \S'   SS\4U 4S jjjr\	   SS\S-  S\
S   S\S-  S	\S
\4   4S jj5       r\R                  " 5       \S 5       5       rSrU =r$ )GraniteMoeSharedRotaryEmbeddingi  inv_freqNr<   c                   > [         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defaultr>  F)
persistentoriginal_inv_freq)rA   rB   max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenr<   rope_parametersr@  compute_default_rope_parametersr   attention_scalingregister_bufferclone)rL   r<   r   rope_init_fnr>  rM   s        r7   rB   (GraniteMoeSharedRotaryEmbedding.__init__  s    "("@"@$*$B$B!44[A!%!E!E>>Y&.t~~>L+7V+L((ZeD0(..2BuUr6   r   ztorch.deviceseq_lenrP   ztorch.Tensorc           	         U R                   S   n[        U SS5      =(       d    U R                  U R                  -  nSnSU[        R
                  " SUS[        R                  S9R                  U[        R                  S9U-  -  -  nXe4$ )	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_thetar   Ng      ?r   r>   ri   )r   ri   )	rG  r   rC   r   r0   arangeint64rj   rw   )r<   r   rN  baserT   attention_factorr>  s          r7   rH  ?GraniteMoeSharedRotaryEmbedding.compute_default_rope_parameters  s    & %%l3fj$/c63E3EIcIc3c U\\!S!5;;?BB&X]XcXcBdgjjk
 ))r6   c                 L   U R                   S S S 2S 4   R                  5       R                  UR                  S   SS5      R	                  UR
                  5      nUS S 2S S S 24   R                  5       n[        UR
                  R                  [        5      (       a0  UR
                  R                  S:w  a  UR
                  R                  OSn[        USS9   UR                  5       UR                  5       -  R                  SS5      n[        R                  " Xf4SS	9nUR                  5       U R                  -  nUR                  5       U R                  -  n	S S S 5        WR	                  UR                   S
9W	R	                  UR                   S
94$ ! , (       d  f       N@= f)Nr   rR   r    mpscpuF)device_typeenabledr>   rS   rQ  )r>  rw   r   rt   rj   r   r,  typestrr   r   r0   r   r   rI  r   ri   )
rL   r   r  inv_freq_expandedposition_ids_expandedrZ  freqsembr   r   s
             r7   rW   'GraniteMoeSharedRotaryEmbedding.forward  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#)rI  r<   rE  rF  r@  )Nr  )r+   r,   r-   r.   r0   rY   r2   r!   rB   staticmethodr   r3   rs   rw   rH  r;  r   rW   r5   rZ   r[   s   @r7   r=  r=    s    llV5 V V  04+/"*&-*(* t* 
~u$	%	* *: ]]_<  <r6   r=  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$ )GraniteMoeSharedModeli  r<   c           	      *  > [         TU ]  U5        UR                  U l        UR                  U l        [
        R                  " UR                  UR                  U R                  5      U l        [
        R                  " [        UR                  5       Vs/ s H  n[        X5      PM     sn5      U l        [        UR                  UR                  S9U l        [#        US9U l        SU l        UR(                  U l        U R+                  5         g s  snf )Nr  r<   F)rA   rB   pad_token_idpadding_idx
vocab_sizer   	EmbeddingrC   embed_tokens
ModuleListr   num_hidden_layersr  layersr^   r  normr=  
rotary_embgradient_checkpointingembedding_multiplier	post_initr  s      r7   rB   GraniteMoeSharedModel.__init__  s     !.. ++LL):):F<N<NPTP`P`ammNSTZTlTlNmnNm)&<Nmn
 ,F,>,>FDWDWX	9H&+#$*$?$?! 	 os   DN	input_idsr   r  r  inputs_embedsr  r   rP   c           
      ^   US L US L-  (       a  [        S5      eU(       a  Uc  [        U R                  S9nUc  U R                  U5      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	XPR                  -  nUn
U R                  X5      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$ )	Nz:You must specify exactly one of input_ids or inputs_embedsrg  r   r    )r   )r<   rw  r   r  r  )r  r   r  r  r  )last_hidden_stater  )
ValueErrorr   r<   rl  get_seq_lengthr0   rR  rt   r   r   r   rs  rq  ro  rn  rp  r   )rL   rv  r   r  r  rw  r  r   past_seen_tokenscausal_maskrO   r  decoder_layers                r7   rW   GraniteMoeSharedModel.forward*  sT    -t";<YZZ0*$++>O  --i8MCRC^==?de <<(;(;A(>}G[G[\_ooL'11!4L(;;')+%
 &(A(AA% #oomJ![[)H4;;+H+HIM)$7*) /# M J 		-0%++
 	
r6   )rl  rs  rr  ro  rp  ri  rq  rj  )NNNNNN)r+   r,   r-   r.   r!   rB   r   r   r   r0   r1   rY   r   r$  r#  r   r   r   rW   r5   rZ   r[   s   @r7   re  re    s    5 "   .2.204(,26!%5
##d*5
 t+5
 &&-	5

 5
 ((4/5
 $;5
 +,5
 
 5
    5
r6   re  gate_logitsr{   c                    U b  [        U [        5      (       d  g[        U [        5      (       aC  U S   R                  n[        R                  " U  Vs/ s H  oUR                  U5      PM     snSS9n[        R                  R                  R                  WSS9n[        R                  " XrSS9u  p[        R                  R                  R                  X5      n
Uc:  [        R                  " U
R                  5       SS9n[        R                  " USS9nGOUR                  u  pUR                  S   X-  -  nUSSS2SS2SS4   R                  XXU45      R                  SX!5      R                  W5      n[        R                   " U
R                  5       U-  SS9[        R                   " USS9-  nUSSS2SS2S4   R                  XX45      R                  SU5      R                  U5      n[        R                   " UU-  SS9[        R                   " USS9-  n[        R                   " XR#                  S5      -  5      nUU-  $ s  snf )ax  
Computes auxiliary load balancing loss as in Switch Transformer - implemented in Pytorch.

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

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

Returns:
    The auxiliary loss.
Nr   rS   rR   )r,  rs   r   r0   r   rj   r   r   r   r   one_hotrm   rw   rt   r   r   r   r   )r  r{   r   r   compute_device
layer_gateconcatenated_gate_logitsrouting_weightsr   selected_expertsexpert_masktokens_per_expertrouter_prob_per_expert
batch_sizesequence_lengthrn  expert_attention_mask router_per_expert_attention_maskoverall_losss                      r7   load_balancing_loss_funcr  e  s+   : *[%"@"@+u%%$Q..#(99^i-j^iPZmmN.K^i-jpq#r hh))112JPR1SO**_DA((%%--.>LK!JJ{'8'8':B "'O!C&4&:&:#
4::1=*B^_ 4AtT12V&OKXYWR,R	 	 "IIk&7&7&9<Q&QWXY\a\e\e!q]
 
 4At+,V&OQRWR%R	 	) "'?=]+]cd!ehmhqhq,!i
 "
 99.1Q1QRS1TTUL+%%[ .ks   Ic                   T  ^  \ rS rSrSS0rSS0rSS/S/40rS\4U 4S	 jjr\	\
        SS\R                  S
-  S\R                  S
-  S\R                  S
-  S\S
-  S\R                  S
-  S\R                  S
-  S\S
-  S\\R                  -  S\\-  4S jj5       5       rSrU =r$ )GraniteMoeSharedForCausalLMi  zlm_head.weightzmodel.embed_tokens.weightlm_headcolwise_gather_outputrO   r   r<   c                 l  > [         TU ]  U5        [        U5      U l        UR                  U l        [
        R                  " UR                  UR                  SS9U l        UR                  U l	        UR                  U l        UR                  U l        UR                  U l        U R                  5         g )NFr?   )rA   rB   re  r'  rj  r   rH   rC   r  router_aux_loss_coefr   r{   r   logits_scalingrt  rK   s     r7   rB   $GraniteMoeSharedForCausalLM.__init__  s     *62
 ++yy!3!3V5F5FUS$*$?$?!!33#)#=#= $33 	r6   Nrv  r   r  r  rw  labelsoutput_router_logitslogits_to_keeprP   c	           
         Ub  UOU R                   R                  nU R                  " S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XR                   R                  -  nSnUb*  U R                  " UU4SU R                   R                  0U	D6nSnU(       aY  [        U
R                  U R                  U R                  U5      nUb*  XR                  UR!                  UR"                  5      -  -  n[%        UUUU
R&                  U
R(                  U
R*                  U
R                  S9$ )a  
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
    Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
    config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
    (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.

Example:

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

>>> model = GraniteMoeSharedForCausalLM.from_pretrained("ibm/PowerMoE-3b")
>>> tokenizer = AutoTokenizer.from_pretrained("ibm/PowerMoE-3b")

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

>>> # Generate
>>> generate_ids = model.generate(inputs.input_ids, max_length=30)
>>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
"Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
```N)rv  r   r  r  rw  rj  )lossaux_lossr   r  rO   r(  router_logitsr*   )r<   r  r'  ry  r,  r3   slicer  r  loss_functionrj  r  r  r{   r   r  rj   r   r   r  rO   r(  )rL   rv  r   r  r  rw  r  r  r  r   outputsrO   slice_indicesr   r  r  s                   r7   rW   #GraniteMoeSharedForCausalLM.forward  sw   J %9$D $++JjJj 	 ** 
)%+'
 
  118B>SV8W8W~ot4]kmA}a,?@A++444%%  ;;11 	D /%%  ((	H !11HKK4LLL(#33!//))!//
 	
r6   )r  r  r'  r{   r   r  rj  )NNNNNNNr   )r+   r,   r-   r.   _tied_weights_keys_tp_plan_pp_planr!   rB   r   r   r0   r1   rY   r   r$  r#  r3   rs   r   rW   r5   rZ   r[   s   @r7   r  r    s    *,GH23H_-z:;H5   .2.204(,26*.,0-.Q
##d*Q
 t+Q
 &&-	Q

 Q
 ((4/Q
   4'Q
 #TkQ
 ell*Q
 
*	*Q
  Q
r6   r  )r  re  r&  )r    )r  )Nr>   N)Icollections.abcr   typingr   r   r0   r   torch.nnr   r    r	   r-  activationsr
   cache_utilsr   r   
generationr   integrationsr   r   r   masking_utilsr   modeling_layersr   modeling_outputsr   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   utils.genericr   r   r   utils.output_capturingr   configuration_granitemoesharedr!   r#   Moduler:   r^   ry   r   r   r   r   rY   r3   r   rw   r   r   r  r&  r=  re  rs   r  r  __all__r*   r6   r7   <module>r     s  * % &   $ & ! . ) f f / 9 Q K F & 7 Y Y 5 B)5 0")) 4 Y'Jbii J (J(*bii *Z.S .Sb,")) ,^( *+ ,2	UU\\ 	U# 	U%,, 	U& %II%<<% 
% <<	%
 LL4'% % % '(%2 )*@)		 @) +@)F0#= 0f Uo U U.><bii ><B J
; J
 J
^ #
*.	O&ell 33d:O&tO& LL4'	O&
 \\CO&d e
"A? e
 e
P fr6   