
    <hJ                       S SK JrJrJrJr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JrJr  SSK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"  SSK#J$r$J%r%J&r&J'r'  SSK(J)r)J*r*  SSK+J,r,  \*" 5       (       a  S SK-J.r.  S SK/J0r0J1r1  OSr.\)" 5       (       a	  S SK2J3r3J4r4  OSu  r4r3\&" 5       (       a  S SK5J6r6  SSK7J8r8  \'Rr                  " \:5      r;S r<SPS jr=S\R|                  S\?S\R|                  4S jr@ SQS\R                  S\R|                  S \R|                  S!\R|                  S"\\R|                     S#\BS$\B4S% jjrC " S& S'\R                  5      rD " S( S)\5      rES*\R|                  S+\?4S, jrFS- rGS. rH\I" \.\3\445      rJS/ rK " S0 S1\R                  5      rL " S2 S3\R                  R                  5      rM " S4 S5\R                  5      rN " S6 S7\S8S99rO " S: S;\R                  5      rP " S< S=\R                  5      rQ " S> S?\R                  5      rR " S@ SA\R                  5      rS " SB SC\5      rT\$ " SD SE\ 5      5       rU " SF SG\R                  5      rV\$ " SH SI\U5      5       rW   SRSJ\\R|                  \X\R|                     S4   SK\\?   S"\\R|                     S\\R|                  \?4   4SL jjrY " SM SN\U\5      rZ/ SOQr[g)S    )AnyCallableOptional	TypedDictUnionN)nn)ACT2FN   )CacheDynamicCacheDynamicLayer)GenerationMixin)AttentionMaskConverter)GradientCheckpointingLayer)BaseModelOutputWithPastMoeCausalLMOutputWithPastMoeModelOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)auto_docstringcan_return_tupleis_torch_flex_attn_availablelogging)is_causal_conv1d_availableis_mamba_2_ssm_available   )GraniteMoeHybridConfig)selective_state_update)mamba_chunk_scan_combined mamba_split_conv1d_scan_combined)causal_conv1d_fncausal_conv1d_updateNN)	BlockMask)make_flex_block_causal_maskc                     U SSU R                   S   S-  24   nU SU R                   S   S-  S24   n[        R                  " U* U4SS9$ )z*Rotates half the hidden dims of the input..N   dim)shapetorchcat)xx1x2s      v/var/www/html/shao/venv/lib/python3.13/site-packages/transformers/models/granitemoehybrid/modeling_granitemoehybrid.pyrotate_halfr5   @   sZ    	
3"!''"+"""	#B	
3q ""	#B99rc2YB''    c                     UR                  U5      nUR                  U5      nX-  [        U 5      U-  -   nX-  [        U5      U-  -   nXg4$ )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.
    position_ids (`torch.Tensor`, *optional*):
        Deprecated and unused.
    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.
)	unsqueezer5   )qkcossinposition_idsunsqueeze_dimq_embedk_embeds           r4   apply_rotary_pos_embrA   G   sS    ( --
&C
--
&Cw;q>C/0Gw;q>C/0Gr6   hidden_statesn_repreturnc                     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)r.   expandreshape)rB   rC   batchnum_key_value_headsslenhead_dims         r4   	repeat_kvrL   b   s_    
 2?1D1D.Ez!!Qa"23::5W\dlmM  e(CTTTr6   modulequerykeyvalueattention_maskscalingdropoutc                 @   [        X R                  5      n[        X0R                  5      n	[        R                  " XR	                  SS5      5      U-  n
Ub"  US S 2S S 2S S 2S UR
                  S   24   n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$ )Nr+   r
   r*   )r-   dtype)ptrainingr   )rL   num_key_value_groupsr/   matmul	transposer.   r   
functionalsoftmaxfloat32torV   rS   rX   
contiguous)rM   rN   rO   rP   rQ   rR   rS   kwargs
key_statesvalue_statesattn_weightscausal_maskattn_outputs                r4   eager_attention_forwardrg   n   s     3 ; ;<JU$?$?@L<<';';Aq'ABWLL!$Q1.D
0@0@0D.D%DE#1 ==((2U]](SVVW\WbWbcL==((6??([L,,|:K''1-88:K$$r6   c                   l  ^  \ rS rSrSrS\S\4U 4S jjr      SS\R                  S\
\R                     S\
\R                     S	\
\   S
\S\
\R                     S\
\\R                  \R                  4      S\\R                  \
\R                     \
\\R                        4   4S jjrSrU =r$ )GraniteMoeHybridAttention   z=Multi-headed attention from 'Attention Is All You Need' paperconfig	layer_idxc                 z  > [         TU ]  5         Xl        X l        Uc-  [        R                  SU R                  R                   S35        UR                  U l        UR                  U l	        UR                  U l        U R                  U R                  -  U l        UR                  U l        U R                  U R                  -  U l        SU l        UR                   U l        U R                  U R                  -  U R                  :w  a&  [%        SU R                   SU R                   S35      e[&        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*                  S9U l        g )NzInstantiating z without passing a `layer_idx` is not recommended and will lead to errors during the forward call if caching is used. Please make sure to provide a `layer_idx` when creating this class.Tz?hidden_size must be divisible by num_heads (got `hidden_size`: z and `num_heads`: z).bias)super__init__rk   rl   loggerwarning_once	__class____name__attention_dropouthidden_sizenum_attention_heads	num_headsrK   rI   rY   	is_causalattention_multiplierrR   
ValueErrorr   Linearattention_biasq_projk_projv_projo_projselfrk   rl   rt   s      r4   rq   "GraniteMoeHybridAttention.__init__   s   " !8!8 9 :, , "(!9!9!--33((DNN:#)#=#= $(NNd6N6N$N!22MMDNN*t/?/??QRVRbRbQc$T^^$4B8 
 ii 0 0$..4==2PW]WlWlmii 0 0$2J2JT]]2Zagavavwii 0 0$2J2JT]]2Zagavavwii 0 0$2B2BI^I^_r6   rB   rQ   r=   past_key_value	use_cachecache_positionposition_embeddingsrD   c                    UR                  5       u  pnU R                  U5      nU R                  U5      nU R                  U5      nUR	                  XU R
                  U R                  5      R                  SS5      nUR	                  XU R                  U R                  5      R                  SS5      nUR	                  XU R                  U R                  5      R                  SS5      nUb  UOSu  nnUb  [        XUU5      u  pUb$  UXS.nUR                  XU R                  U5      u  p[        nU R                  R                  S:w  a  [        U R                  R                     nU" U UUUU4U R                   (       d  SOU R"                  U R$                  S.UD6u  nnUR	                  XS5      nU R'                  U5      nUU4$ )	Nr   r+   r&   )r<   r;   r   eager        )rS   rR   r*   )sizer   r   r   viewry   rK   r[   rI   rA   updaterl   rg   rk   _attn_implementationr   rX   rv   rR   r   )r   rB   rQ   r=   r   r   r   r   ra   bszq_len_query_statesrb   rc   r;   r<   cache_kwargsattention_interfacerf   rd   s                        r4   forward!GraniteMoeHybridAttention.forward   s    &**,A{{=1[[/
{{=1#((T^^T]]S]]^_abc__S1I1I4==Yccdeghi
#((T5M5Mt}}]gghiklm*=*I&|S*';LVY[^'_$L%#&sUL'5'<'<ZW[WeWegs't$J(?;;++w6"9$++:Z:Z"[$7	%
  $}}C$2H2HLL	%
 	%
!\ "&&s26kk+.L((r6   )rv   rk   rK   rw   rz   r   rl   ry   rY   rI   r   r   rR   r   )NNNFNN)ru   
__module____qualname____firstlineno____doc__r    intrq   r/   Tensorr   
LongTensorr   booltupler   __static_attributes____classcell__rt   s   @r4   ri   ri      s    G`5 `# `F 2637*.59KO0)||0) !.0) u//0	0)
 !0) 0) !!1!120) &eELL%,,,F&GH0) 
u||Xell3XeELL>Q5RR	S0) 0)r6   ri   c                     ^  \ rS rSrSrSrSrSr\R                  S4S\
4U 4S jjjr SS\R                  S\R                  S	\S
\\\\4      S\\R                  \R                  4   4
S jjrS\R(                  4S jrSS	\\   S\4S jjrS\\\R                     \\R                     4   4S jr\SS\\\\R2                           SS4S jj5       rSrU =r$ ) HybridMambaAttentionDynamicCache   a|  
A dynamic cache that can handle both the attention cache (which has a seq_len dimension) and the mamba cache
(which has a constant shape regardless of seq_len).

This cache has two sets of lists of tensors: `key_cache` and `value_cache` for attention cache and `conv_states`
and `ssm_states` for mamba cache. Each of these lists has `num_layers` tensors. The expected shape for each tensor
For attention layers, `key_cache` and `value_cache` have a shape of `(batch_size, num_heads, seq_len, head_dim)`,
while `conv_states` and `ssm_states` have a shape of `(batch_size, 0)` (empty tensors).
For mamba layers, `key_cache` and `value_cache` have a shape of `(batch_size, 0)` (empty tensors),
while `conv_states` represents the convolution state and has a shape of `(batch_size, d_inner, d_conv)`,
and `ssm_states` represents the ssm state and has a shape of `(batch_size, d_inner, d_state)`.
NFrk   c                 H  > [         T	U ]  [        S9  UR                  U l        SU l        UR
                  nUR                  n/ U l        / U l        / U l	        [        UR                  5       GH%  nU R                  U   S:X  a  U =R                  [        R                  " UUR                  UR                  -  SUR                   -  U-  -   UUUS9/-  sl        U =R                  [        R                  " UUR"                  UR$                  UUUS9/-  sl        M  U =R                  [        R&                  " / /U-  US9/-  sl        U =R                  [        R&                  " / /U-  US9/-  sl        U R                  R)                  U5        GM(     [        UR                  5       Vs/ sH  n[        R&                  " / /U-  US9PM     snU l        [        UR                  5       Vs/ sH  n[        R&                  " / /U-  US9PM     snU l        g s  snf s  snf )N)layer_classesFmambar+   devicerV   r   )rp   rq   r   layers_block_typehas_previous_statemamba_d_convmamba_d_stateconv_states
ssm_statestransformer_layersrangenum_hidden_layersr/   zerosmamba_expandrw   mamba_n_groupsmamba_n_headsmamba_d_headtensorappend	key_cachevalue_cache)
r   rk   
batch_sizerV   r   conv_kernel_sizessm_state_sizeir   rt   s
            r4   rq   )HybridMambaAttentionDynamicCache.__init__   s   |4!'!9!9"'!..--"$v//0A%%a(G3  KK",,v/A/AAAH]H]D]`nDnn(%#%   KK",,++&%#	$ 	   U\\2$2CF%S$TT ELL"
1B6$R#SS''..q11 14 SXX^XpXpRqrRqQ%,,tj'8HRqrTYZ`ZrZrTstTsqELL"
):6JTst sts   ."H/"Hrb   rc   rl   r   rD   c                 |   U R                   U   R                  S   S:X  a  XR                   U'   X R                  U'   Ob[        R                  " U R                   U   U/SS9U R                   U'   [        R                  " U R                  U   U/SS9U R                  U'   U R                   U   U R                  U   4$ )Nr*   r   r+   r,   )r   r.   r   r/   r0   )r   rb   rc   rl   r   s        r4   r   'HybridMambaAttentionDynamicCache.update  s     >>)$**2.!3(2NN9%*6Y'(-		4>>)3Lj2Y_`(aDNN9%*/))T5E5Ei5PR^4_ef*gDY'~~i($*:*:9*EEEr6   beam_idxc                    [        [        U R                  5      5       GHT  nU R                  U   R                  nU R                  U   R	                  SUR                  U5      5      U R                  U'   U R                  U   R                  nU R                  U   R	                  SUR                  U5      5      U R                  U'   U R                  U   R                  nU R                  U   R	                  SUR                  U5      5      U R                  U'   U R                  U   R                  nU R                  U   R	                  SUR                  U5      5      U R                  U'   GMW     g)zDReorders the cache for beam search, given the selected beam indices.r   N)	r   lenr   r   index_selectr_   r   r   r   )r   r   rl   r   s       r4   reorder_cache.HybridMambaAttentionDynamicCache.reorder_cache+  s=   s4>>23I^^I.55F(,y(A(N(NqRZR]R]^dRe(fDNN9%%%i077F*.*:*:9*E*R*RSTV^VaVabhVi*jDY'%%i077F*.*:*:9*E*R*RSTV^VaVabhVi*jDY'__Y/66F)-)C)P)PQRT\T_T_`fTg)hDOOI& 4r6   c                     XR                   ;  a  U R                   S   OUn[        U R                  5      U::  a  gU R                  U   R                  S   $ )zYReturns the sequence length of the cached states. A layer index can be optionally passed.r   rU   )r   r   r   r.   )r   rl   s     r4   get_seq_length/HybridMambaAttentionDynamicCache.get_seq_length8  sP     3<CZCZ2ZD++A.`i	t~~)+~~i(..r22r6   c                     [        S5      eNzIHybridMambaAttentionDynamicCache does not have a legacy cache equivalent.NotImplementedErrorr   s    r4   to_legacy_cache0HybridMambaAttentionDynamicCache.to_legacy_cache@  s    !"mnnr6   past_key_valuesr   c                     [        S5      er   r   )clsr   s     r4   from_legacy_cache2HybridMambaAttentionDynamicCache.from_legacy_cacheC  s    !"mnnr6   )r   r   r   r   r   r   r   N)r   )ru   r   r   r   r   r   r   is_compileabler/   float16r    rq   r   r   r   dictstrr   r   r   r   r   r   r   classmethodFloatTensorr   r   r   r   s   @r4   r   r      s1    IKNIN_c %u5 %u %uX 26FLLF llF 	F
 tCH~.F 
u||U\\)	*F"ie&6&6 i3 3c 3ouU\\':E%,,<O'O!P o ouUEVEV?W9X0Y oes o or6   r   input_tensorpad_sizec                     [        U R                  5      S:X  a
  SSSSSUSS4OSSSUSS4n[        R                  R                  R                  XSSS9$ )zv
Padding x tensor with `pad_size` on the seq_len dim (dim=1)

Assumes that we only have tensors of either size 4 or 3
   r   constant)moderP   )r   r.   r/   r   r\   pad)r   r   	pad_shapes      r4   pad_tensor_by_sizer   K  sd     47|7I7I3Ja3OAq!Q!Q/VWYZ\]_gijlmUnI88""<ST"UUr6   c                    [        X5      n [        U R                  5      S:X  a-  U R                  U R                  S   SX R                  S   5      $ U R                  U R                  S   SX R                  S   U R                  S   5      $ )z
Padding input_tensor with `pad_size` on the seq_len dim (dim=1) and
simultaneously splitting it into chunk sequences.

Assumes that we only have tensors of either size 4 or 3
r
   r   r*   r+   )r   r   r.   rG   )r   r   
chunk_sizes      r4   reshape_into_chunksr   V  s     &l=L
<!###L$6$6q$92zK]K]^_K`aa ##q!2z3E3Ea3H,J\J\]^J_
 	
r6   c           	      
   U R                  S5      nU S   R                  " / U R                  5       QUP76 n [        R                  " [        R                  " XU R
                  [        R                  S9SS9nU R                  U) S5      n [        R                  " U SS9n[        R                  " [        R                  " XU R
                  [        R                  S9SS9nUR                  U) [        R                  * 5      nU$ )zg
More stable segment sum calculation. Uses cumulative sums and masking instead of direct subtractions.
r*   .Nr   diagonalr   rU   r,   )
r   rF   r/   trilonesr   r   masked_fillcumsuminf)r   r   masktensor_segsums       r4   segment_sumr   j  s     ""2&J  	*11S<3D3D3FS
SL::ejj@S@S[`[e[efqstD++TE15LLL26M ::ejj@S@S[`[e[efqrsD!--teeiiZ@Mr6   c                     UbO  UR                   S   S:  a<  UR                   S   S:  a)  U R                  nXSS2SS2S4   -  R                  U5      n U $ )ze
Tunes out the hidden states for padding tokens, see https://github.com/state-spaces/mamba/issues/66
Nr   r   )r.   rV   r_   )rB   rQ   rV   s      r4   apply_mask_to_padding_statesr     s_     !n&:&:1&=&AnFZFZ[\F]`aFa##&1d
)CCGGNr6   c                     ^  \ rS rSrSrS\S\4U 4S jjr    SS\R                  S\
\   S\
\R                     S	\
\R                     S
\
\R                     4
S jjr   SS\
\   S\
\R                     S	\
\R                     4S jjr    SS\
\   S\
\R                     S	\
\R                     S
\
\R                     4S jjrSrU =r$ )GraniteMoeHybridMambaLayeri  u'  
Compute ∆, A, B, C, and D the state space parameters and compute the `contextualized_states`.
A, D are input independent (see Mamba paper [1] Section 3.5.2 "Interpretation of A" for why A isn't selective)
∆, B, C are input-dependent (this is a key difference between Mamba and the linear time invariant S4,
and is why Mamba is called **selective** state spaces)

The are a few differences between this and Mamba2Mixer:
- The variable use_precomputed_states is slightly different due to the HybridCache structure
- There's a few non-obvious bugs fixed with batching in the slow path that exist in main
- Some extra variables that our layer doesn't need have been removed
- We ported most of the refactors in https://github.com/huggingface/transformers/pull/35154, which is (as of Dec 18, 2024) unmerged
rk   rl   c           	        > [         TU ]  5         UR                  U l        UR                  U l        UR
                  U l        UR                  U l        [        UR                  U R                  -  5      U l        X 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R2                  U l        UR6                  U l        S[;        S5      4U l        SU l        SU l         U R                  SU R0                  -  U R                  -  -   U l!        [D        RF                  " U RB                  U RB                  UR                  U R                  U RB                  U R                  S-
  S9U l$        U R                  U RB                  -   U R                  -   n[D        RJ                  " U R                  UU R(                  S9U l&        [D        RN                  " [P        RR                  " U R                  5      5      U l*        [P        RV                  " SU R                  S-   5      n[D        RN                  " [P        RX                  " U5      5      U l-        S	U RZ                  l.        [_        U R                  U R,                  S
9U l0        [D        RN                  " [P        RR                  " U R                  5      5      U l1        S	U Rb                  l.        [D        RJ                  " U R                  U R                  U R(                  S9U l2        [f        (       d  [h        Rk                  S5        g [h        Rk                  S5        g )Nr   r   gMbP?g?r+   r   )in_channelsout_channelsro   kernel_sizegroupspaddingrn   Tepsa  The fast path is not available because on of `(selective_state_update, causal_conv1d_fn, causal_conv1d_update)` is None. Falling back to the naive implementation. To install follow https://github.com/state-spaces/mamba/#installation and https://github.com/Dao-AILab/causal-conv1dzOThe fast path for GraniteMoeHybrid will be used when running the model on a GPU)6rp   rq   r   ry   rw   r   r   r   r   r   r   intermediate_sizerl   mamba_conv_biasuse_conv_bias
hidden_act
activationr	   actmamba_proj_biasuse_biasrms_norm_epslayer_norm_epsilonr   n_groupsr   rK   mamba_chunk_sizer   floattime_step_limittime_step_mintime_step_maxconv_dimr   Conv1dconv1dr}   in_proj	Parameterr/   r   dt_biasarangelogA_log_no_weight_decayGraniteMoeHybridRMSNormGatednormDout_projis_fast_path_availablerr   rs   )r   rk   rl   projection_sizeArt   s        r4   rq   #GraniteMoeHybridMambaLayer.__init__  s   --!--$22 & 3 3!$V%8%84;K;K%K!L"#33 ++&++,.."("5"5--++ 11 !$U5\2" ..T]]1BTEXEX1XXii''--==))A-
 004==@4>>Qyy
 ||EJJt~~$>? LLDNNQ./\\%))A,/
&*

#01G1GTMdMde	ejj89"&		$"8"8$:J:JQUQ^Q^_%%>  qrr6   rB   cache_paramsr   rQ   seq_idxc                 h   [        X5      nU R                  U5      nUR                  u  pxn	U R                  U R                  -  n
US L=(       a    UR
                  =(       a    US:H  =(       aw    UR                  U R                     R                  S   UR                  U R                     R                  S   s=:H  =(       a    U:H  Os  =(       a    US L=(       a    US   S:  nU(       Ga  UR                  S5      R                  U R                  U R                  U R                  /SS9u  pn[        UUR                  U R                     U R                  R                   R                  S5      U R                  R"                  U R$                  5      n[&        R                  " UU R                  X/SS9u  pn[&        R(                  " U R*                  R-                  5       5      * nUS S 2S S4   S S 2S S 2S 4   R/                  SU R0                  U R                  5      R3                  [&        R4                  S9nUS S 2S S 2S 4   R/                  SSU R0                  5      nU R6                  S S 2S S4   R/                  SU R0                  5      nU R8                  S S 2S S4   R/                  SU R0                  5      nUR;                  XpR                  UR                  S   U R                  -  5      nUR;                  XpR                  UR                  S   U R                  -  5      nUR;                  XpR                  U R0                  5      n[=        UR                  U R                     UUUUUUS USS9
nUR;                  XpR                  U R0                  -  5      nU R?                  X5      nU RA                  U5      S S 2S S4   nU$ [&        R(                  " U R*                  R-                  5       5      * nU RB                  S	[-        S
5      4:X  a  0 OSU RB                  0nU RD                  (       a  Uc  [G        UU R                  R                   R                  S5      U R                  R"                  U R6                  U4U R8                  U RH                  UU R$                  U R>                  R                   U R>                  RJ                  U R@                  R                   U R@                  R"                  U R0                  U R                  SSS.UD6nU$ UR                  U R                  U R                  U R                  /SS9u  pnUbv  URM                  SS5      n[N        RP                  RS                  UU RT                  UR                  S   -
  S45      nUR                  U R                     RW                  U5        U R$                  S;  aH  U RY                  U R                  URM                  SS5      5      SS U24   RM                  SS5      5      nOn[[        URM                  SS5      U R                  R                   R                  S5      U R                  R"                  U R$                  US9RM                  SS5      n[        X5      n[&        R                  " UU R                  X/SS9u  pn[]        UR;                  XxSU R0                  5      UUUR;                  XxU R                  S5      UR;                  XxU R                  S5      4U RH                  U R8                  S USU R6                  SS.UD6u  nnUb+  Ub(  UR                  U R                     RW                  U5        UR;                  XxS5      nU R?                  UU5      nU RA                  U5      nU$ )Nr   r   r*   r,   .rV   T)zr  dt_softplusr   r   dt_limitF)r#  r   r*  r  rmsnorm_weightrmsnorm_epsoutproj_weightoutproj_biasheaddimngroupsnorm_before_gatereturn_final_statesr+   )siluswish)r1   weightro   r  r*  )r   r#  r-  r*  r7  r  r.  )/r   r  r.   r  r   r   r   rl   r   squeezesplitr  r  ry   r%   r  r:  ro   r  r/   expr  r  rF   rK   r_   r^   r  r#  r   r!   r"  r$  r  rX   r#   r   variance_epsilonr[   r   r\   r   r   copy_r  r$   r"   )r   rB   r)  r   rQ   r*  projected_statesr   seq_lenr   groups_time_state_sizeuse_precomputed_statesgatehidden_states_B_CdtBCr'  r  r#  hidden_states_reshapedoutdt_limit_kwargshidden_states_B_C_transposedr   scan_output	ssm_states                              r4   cuda_kernels_forward/GraniteMoeHybridMambaLayer.cuda_kernels_forward  s    5]S<<6 "/!4!4
Q!%1D1D!D $ &//&1& ((8>>qA&&t~~6<<Q? & d*& q!A% 	 "*:*B*B1*E*K*K''GR +L +'DR
 !5!((8""**1-  ! #(++!'')?X#Ma 4::++-..A!T3,1d
+222t}}dFYFYZ]]didqdq]rAAq$J&&r2t}}=Bll1dC<077DMMJGq$|$++B>Az==!''!*2MNAz==!''!*2MNA%2%7%7
NNTXTaTa%b"2''7& M *..z>>DMM;YZM IIm:M --.q$|<C| 
w 4::++-..A$($8$8S%,<O$ObV`bfbvbvUwO }}!56$KK&&..q1KK$$LL ff####'99#3#3 $		 : :#'==#7#7!%!3!3 MM MM%*(-#$ &%l 
A /?.D.D++T]]DNNKQS /E /+  + 4E3N3NqRS3T0"$--"3"34..1M1S1STV1WWYZ[#K !,,T^^<BB;O??*;;(,$5$?$?1$EFsHWH}U__`acde)% )9+55a;#{{1199!<![[--#'?? ')  i1o & %AAR$c!&+kk%++-C\'#! *C!&&zBNFF:rBFF:rB*  $ff#(, LL $* &*&Y" (\-E ++DNN;AA)L)..zBG"iiT: mmK0
r6   c                    UR                   u  pVnUR                  n[        X5      nU R                  U5      n	U	R	                  U R
                  U R                  U R                  /SS9u  pnUS L=(       a    UR                  =(       a    US:H  =(       aw    UR                  U R                     R                   S   UR                  U R                     R                   S   s=:H  =(       a    U:H  Os  =(       a    US L=(       a    US   S:  nU(       GaT  UR                  U R                     R                  SSS9UR                  U R                  '   US S 2SS S 24   R                  UR                  U R                     R                  5      UR                  U R                     S S 2S S 2S4'   UR                  U R                     R                  U R                  R                   R                  S9n["        R$                  " XR                  R                   R'                  S5      -  SS9nU R(                  (       a  XR                  R*                  -   nU R-                  U5      nOUbu  UR/                  SS5      n[0        R2                  R5                  XR6                  UR                   S   -
  S45      nUR                  U R                     R9                  U5        U R-                  U R                  UR/                  SS5      5      SS U24   R/                  SS5      5      n[        X5      n["        R                  " UU R
                  U R:                  U R<                  -  U R:                  U R<                  -  /SS9u  nnn["        R>                  " U R@                  RC                  5       5      * nU(       Ga  UR                  U R                     R                  nUS S 2SS S 24   S S 2S S4   nUR/                  SS5      RE                  X\R                   S   U RF                  5      nU RH                  S	   RE                  U RH                  R                   S   U RF                  5      n["        R0                  R2                  RK                  UUR                  UR                  5      -   5      n["        RL                  " XRN                  S   U RN                  S   5      nUS
   RE                  U R                  U RF                  U R<                  5      R                  ["        RP                  S9n["        R>                  " US	   U-  5      R                  US9nURS                  XPR:                  S5      SS S S 24   nURE                  XPR:                  U R                  U R:                  -  UR                   S   5      RU                  5       nURS                  USUR                   S   5      nUS	   USS S S 24   -  nURS                  USU RF                  5      nUUS	   -  R                  US9nUR                  U R                     R9                  UR                  U R                     U-  U-   5        URS                  XPR:                  S5      SS S S 24   nURE                  XPR:                  U R                  U R:                  -  UR                   S   5      RU                  5       nURS                  USUR                   S   5      nUR                  U R                     R                  UR                  UR                  S9nURW                  XPR                  -  U RF                  U R<                  5      nURW                  XPR                  -  U R<                  S5      n["        RX                  " UU5      nURW                  XPR                  U RF                  5      nU RZ                  S	   RE                  U RZ                  R                   S   U RF                  5      nUUU-  -   R                  UR                  5      nURS                  US5      S S 2S S4   nGO[0        R2                  RK                  XRH                  -   5      n["        RL                  " XRN                  S   U RN                  S   5      nURS                  XVSU RF                  5      RC                  5       nURS                  XVSU R<                  5      RC                  5       nURS                  XVSU R<                  5      RC                  5       nUR]                  U R                  U R:                  -  SU R                  S9nUR]                  U R                  U R:                  -  SU R                  S9nU R^                  X`R^                  -  -
  U R^                  -  nU RZ                  S	   [a        UU5      -  nUUS	   -  nUR                  UR                  5      U-  nUUUU4 V s/ sH  n [c        U UU R^                  5      PM     sn u  nnnnURe                  SSSS5      n["        Rf                  " USS9n!["        R>                  " [i        U5      5      n"US S 2S S 2S S 2S S S 2S S 24   US S 2S S 2S S S 2S S 2S S 24   -  n#U#R%                  SS9n$U$S	   U"Re                  SSSSS5      S	   -  n%U%R%                  SS9n&U&S	   US S 2S S 2S 4   -  R%                  SS9n'["        R>                  " U!S S 2S S 2S S 2SS 24   U!-
  5      n(UU(Re                  SSSS5      S	   -  n)U)SS S S 24   US	   -  R%                  SS9n*U(       a9  UR                  U R                     S S 2S S4   R                  U*R                  S9n+O["        Rj                  " U*S S 2S S24   5      n+["        Rl                  " U+U*/SS9n*["        R>                  " [i        [0        R2                  R5                  U!S S 2S S 2S S 2S4   S5      5      5      n,U,R/                  SS5      n,U,S
   U*S S 2S S 2S S4   -  R%                  SS9n-U-S S 2S S24   U-S S 2S4   n.n*["        R>                  " U!5      n/USS S S 24   U*S S 2S S 2S S4   -  n0U/Re                  SSSS5      n1U0R%                  S5      U1S	   -  n2U'U2-   nURS                  USU R                  U RF                  5      nUU-   nUS:  a  US S 2S U2S S 2S S 24   nURS                  XVS5      nU.b+  Ub(  UR                  U R                     R9                  U.5        U Ro                  UU
5      n3U Rq                  U3R                  U5      5      n4U4$ s  sn f )Nr*   r,   r   r   )shiftsdimsr   r+   .r   ).NNr,  r   )r-   output_sizer
   r   rU   )r   r   )9r.   rV   r   r  r<  r  r  ry   r   r   rl   r   rollr_   r   r  r:  r/   sumr;  r	  ro   r  r[   r   r\   r   r   r?  r  r   r=  r  r  rF   rK   r  softplusclampr  r^   rG   r`   r   bmmr#  repeat_interleaver   r   r   permuter   r   
zeros_liker0   r"  r$  )5r   input_statesr)  r   rQ   r   rA  r   rV   r@  rD  rE  rF  rC  r   rL  rB   rG  rH  r'  cache_devicer  dAdBdBxr   ssm_states_reshaped
C_reshapedyr#  r   
D_residualtA_cumsumLG_intermediateGM_intermediateMY_diagdecay_statesB_decaystatesprevious_statesdecay_chunk
new_statesrN  state_decay_outC_times_statesstate_decay_out_permutedY_offrM  contextualized_statess5                                                        r4   torch_forward(GraniteMoeHybridMambaLayer.torch_forward  s1    ".!3!3
Q"" 4LQ<<5&6&<&<''GR '= '
#
 $ &//&1& ((8>>qA&&t~~6<<Q? & d*& q!A% 	 "7C7O7OPTP^P^7_7d7dlnuw7d7xL$$T^^4ARSTVWYZSZA[A^A^_k_w_wx|  yG  yG  `H  `O  `O  BPL$$T^^4Q2X> '224>>BEET[[M_M_MfMfEgK %		kk0088;;! !!$58H8H$H! $): ; '/@/J/J1a/P, mm//03H3HKgKmKmnpKq3qst2u ((8>>{K $5F5P5PQRTU5V)WX[]e^e]eXe)f)p)pqrtu)v w89J[#kk##T]]T5H5H%H$--Z^ZmZmJmn
q! YYtzz'')**!'224>>BIIL Aq!GQc\*Ba#**:xx|T]]SBll9-44T\\5G5G5JDMMZG$$--b7::bhh3G.GHBR!5!5a!8$:N:Nq:QRB/"))$..$--I\I\]``glgtgt`uA))ByMA-.22,2GB
 		*mmR8dAFA]]DNNdmm4SUVU\U\]_U`allnA		*b!''"+6AI3a<0B *11*b$--PMi0044L4IC ##DNN399''7"<sB 		*mmR8dAFA]]DNNdmm4SUVU\U\]_U`allnA		*b!''"+6A &00@CC188[\[b[bCcJ",//*~~2Mt}}^b^q^q"r
^^ ;T=P=PRSTJ		-z:Az>>4==AA y!((a$--HA]Q&&**1773A 		*b)!T3,7A ''\\(9:BR!5!5a!8$:N:Nq:QRB)11*r4==Y__aM		*r43F3FGMMOA		*r43F3FGMMOA##DNNdmm$CX\XfXf#gA##DNNdmm$CX\XfXf#gA'OO*CCtVH	*-?x-XXJ *ByM9M](()B.A cpqrtuwxay%zay\]&9!Xt&Way%z"M1a 		!Q1%A||A2.H 		+a.)A q!Qa23a1dAq!8K6LLN""r"*A y\AIIaAq!,DY,OON""r"*A 	l]1a:%>>CCCJF !99XaArsl%;h%FGL,..q"b!<YGGGc4l+mI.FFKKPQKRF &"."9"9$.."I!TSV,"W"Z"Zbhbobo"Z"p"'"2"26!RaR%="AYY8a@F))K0A0A(1aQRTV;BWY_0`$abK%//15K%o61dC9PPUUZ[U\J *1crc6 2Jq"u4EIF $ii1OT1oq!T30GGN'6'>'>q!Q'J$#''+.Fy.QQE A		*b$..$--HAJA!|a'1a'(		*r2A $)A''7==iHii4(
 !%knnU.C D$$G &{s    u?c                 ~   [         (       aA  SU R                  R                  R                  R                  ;   a  U R                  XX4U5      $ Ub  [        S5      eUR                  nUbC  UR                  S   S:  a0  UR                  S   S:  a  XS S 2S S 2S 4   -  R                  U5      nU R                  XX45      $ )Ncudaz\`seq_idx` support requires fast path support. Please install `mamba_ssm` and `causal_conv1d`r   r   )r%  r  r:  r   typerO  r   rV   r.   r_   ry  )r   rB   r)  r   rQ   r*  ra   rV   s           r4   r   "GraniteMoeHybridMambaLayer.forwardW  s     "!f0C0C0J0J0O0O&O,,].jqrr%n  ##%.*>*>q*AA*E.J^J^_`JadeJe*Aq$J-GGKKERM!!-~^^r6   )r  r#  r  r  r   r  r  r   r  rK   rw   r  r  rl   r  r  r"  ry   r$  r   r  r  r  r  r	  )NNNN)NNN)ru   r   r   r   r   r    r   rq   r/   r   r   r   r   	IntTensorrO  ry  r   r   r   r   s   @r4   r   r     sP   As5 As# AsL DH5915-1g||g ?@g !!1!12	g
 !.g %//*gZ DH5915L% ?@L% !!1!12	L%
 !.L%d DH5915-1_ ?@_ !!1!12	_
 !._ %//*_ _r6   r   c                   6   ^  \ rS rSrSU 4S jjrSS jrSrU =r$ )r!  in  c                    > [         TU ]  5         [        R                  " [        R
                  " U5      5      U l        X l        g r   rp   rq   r   r  r/   r   r:  r>  r   rw   r  rt   s      r4   rq   %GraniteMoeHybridRMSNormGated.__init__o  s-    ll5::k#:; #r6   c                    UR                   nUR                  [        R                  5      nUb?  U[        R
                  R                  UR                  [        R                  5      5      -  nUR                  S5      R                  SSS9nU[        R                  " X@R                  -   5      -  nU R                  UR                  U5      -  $ Nr+   r*   T)keepdim)rV   r_   r/   r^   r   r\   r8  powmeanrsqrtr>  r:  )r   rB   rD  input_dtypevariances        r4   r   $GraniteMoeHybridRMSNormGated.forwardt  s    #))%((7)BMM,>,>twwu}}?U,VVM $$Q',,R,>%H?T?T4T(UU{{]--k:::r6   r>  r:  gư>r   )ru   r   r   r   rq   r   r   r   r   s   @r4   r!  r!  n  s    $
	; 	;r6   r!  c                   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$ )	GraniteMoeHybridMLPi  zj
MLP layer for shared experts

Args:
    config:
        Configuration object with model hyperparameters.
rk   c                 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 )Nr+   Frn   )rp   rq   rw   
input_sizeshared_intermediate_sizer	   r
  r  r   r}   input_linearoutput_linearr   rk   rt   s     r4   rq   GraniteMoeHybridMLP.__init__  s     ,,!:: !2!23IIdoot7G7G!7KRWXYYt'7'7uUr6   rB   rD   c                     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+   r*   r,   r   r   )r  chunkr  r  )r   rB   chunked_hidden_statess      r4   r   GraniteMoeHybridMLP.forward  s^    ))-8 - 3 3A2 3 >(=a(@ADYZ[D\\**=9r6   )r  rw   r  r  r  )ru   r   r   r   r   r    rq   r/   r   r   r   r   r   s   @r4   r  r    s7    V5 VU\\ ell  r6   r  c                       \ rS rSr% Sr\R                  \S'   \R                  \S'   \\S'   \\S'   \R                  \S'   Sr
g	)
GraniteFlashAttentionKwargsi  aR  
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.

Attributes:
    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_kr*   N)ru   r   r   r   r   r/   r   __annotations__r   r  r   r  r6   r4   r  r    s7    " ######__r6   r  F)totalc                   8   ^  \ rS rSrSU 4S jjrS rS rSrU =r$ )GraniteMoeHybridRMSNormi  c                    > [         TU ]  5         [        R                  " [        R
                  " U5      5      U l        X l        g)z6
GraniteMoeHybridRMSNorm is equivalent to T5LayerNorm
Nr  r  s      r4   rq    GraniteMoeHybridRMSNorm.__init__  s/     	ll5::k#:; #r6   c                    UR                   nUR                  [        R                  5      nUR	                  S5      R                  SSS9nU[        R                  " X0R                  -   5      -  nU R                  UR                  U5      -  $ r  )	rV   r_   r/   r^   r  r  r  r>  r:  )r   rB   r  r  s       r4   r   GraniteMoeHybridRMSNorm.forward  sw    #))%((7 $$Q',,R,>%H?T?T4T(UU{{]--k:::r6   c                 ^    [        U R                  R                  5       SU R                   3$ )Nz, eps=)r   r:  r.   r>  r   s    r4   
extra_repr"GraniteMoeHybridRMSNorm.extra_repr  s*    ))*+6$2G2G1HIIr6   r  r  )	ru   r   r   r   rq   r   r  r   r   r   s   @r4   r  r    s    $;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$ )
GraniteMoeHybridParallelExpertsi  num_expertsr  rT  rD   Nc                    > [         TU ]  5         [        R                  " [        R
                  " XU5      5      U l        Xl        X l        X0l	        g)a]  
Initialize the GraniteMoeHybridParallelExperts 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)
rp   rq   r   r  r/   emptyr:  r  r  rT  )r   r  r  rT  rt   s       r4   rq   (GraniteMoeHybridParallelExperts.__init__  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 GraniteMoeHybridParallelExperts module.

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

Returns:
    Tensor: Output tensor.
r   r,   )	r<  r   r  r   Flinearr:  r/   r0   )r   inputsexpert_size
input_listoutput_listr   resultss          r4   r   'GraniteMoeHybridParallelExperts.forward  sh     \\+1\5
t''(Aqxx
t{{1~FG )))KQ/r6   )r  r  rT  r:  	ru   r   r   r   r   rq   r   r   r   r   s   @r4   r  r    s.    'C 'S 's 't '. r6   r  c                   >   ^  \ rS rSrS\S\S\4U 4S jjrS rSrU =r$ )GraniteMoeHybridTopKGatingi  r  r  top_kc                 z   > [         TU ]  5         X l        Xl        X0l        [
        R                  " XSS9U l        g)z
Initialize the top-k gating mechanism.
Args:
    input_size (`int`):
        Size of the input.
    num_experts (`int`):
        Number of experts.
    top_k (`int`):
        Number of top experts to select.
Frn   N)rp   rq   r  r  r  r   r}   layer)r   r  r  r  rt   s       r4   rq   #GraniteMoeHybridTopKGating.__init__  s2     	&$
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$ )Nr   r,   r   rV   r   trunc)rounding_mode)r  r  topkr  r/   r]   type_asr   r   r  rV   r   scatterlongrV  tolistflattensortdiv)r   rB   logitstop_k_logitstop_k_indicestop_k_gatesr   gatesr  top_k_expertsr   index_sorted_expertsbatch_indexbatch_gatess                 r4   r   "GraniteMoeHybridTopKGating.forward  s"   M*002&,kk$**!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   )r  r  r  r  r  r   s   @r4   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	$ )GraniteMoeHybridMoEi$  z
A Sparsely gated mixture of experts layer with 1-layer Feed-Forward networks as experts.

Args:
    config:
        Configuration object with model hyperparameters.
rk   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+   )r  r  r  )rp   rq   rw   r  r  r	   r
  r  r  num_local_expertsr  r  r  num_experts_per_tokrouterr  s     r4   rq   GraniteMoeHybridMoE.__init__-  s     ,,!33 !2!23;$$doot7G7G!7K
 =$$d&6&6
 100,,
r6   c                    UR                  5       u  p#nUR                  SU5      nU R                  U5      u  pVpxn	X   n
U R                  X5      nUR	                  SSS9nU R                  US   5      US   -  nU R                  X5      nXSS2S4   -  n[        R                  " X#-  U R                  4UR                  UR                  S9nUR                  SXm5      nUR                  X#U R                  5      nX4$ )z
Forward pass of the mixture of experts layer.

Args:
    layer_input (Tensor):
        Input tensor.

Returns:
    Tensor:
        Output tensor.
    Tensor:
        Router logits.
r*   r+   r,   r   r   Nr  )r   rG   r  r  r  r  r  r/   r   r  rV   r   	index_addr   )r   layer_inputr   lengthemb_sizer   r  r  r  router_logitsexpert_inputsrB   r  expert_outputsr   layer_outputs                   r4   r   GraniteMoeHybridMoE.forward@  s    !, 0 0 2X!))"h7BF++kBZ?-#0))-E - 3 3A2 3 >(=a(@ADYZ[D\\++MG'ag*>>S\4??;>CWCW`n`u`uvq+F#((dooF**r6   )r  rw   r  r  r  r  )
ru   r   r   r   r   r    rq   r   r   r   r   s   @r4   r  r  $  s    
5 
&+ +r6   r  c                   l  ^  \ rS rSrS\S\4U 4S jjr       SS\R                  S\	\R                     S\	\
   S\	\   S	\	\   S
\	\R                     S\	\   S\	\\R                  \R                  4      S\\   S\\R                   \	\\R                   \R                   4      4   4S jjrSrU =r$ )GraniteMoeHybridDecoderLayeri`  rk   rl   c                 (  > [         TU ]  5         UR                  U l        S U l        UR                  S:  a  [        U5      U l        [        UR                  UR                  S9U l	        [        UR                  UR                  S9U l
        UR                  U l        [        U5      U l        S U l        UR                  U   S:X  a  [!        X5      U l        O[#        X5      U l        UR                  U   U l        ['        USS5      S:  U l        g )Nr   r  r   r  )rp   rq   rw   	self_attnr  r  block_sparse_moer  r  input_layernormpost_attention_layernormresidual_multiplierr  
shared_mlpr   r   r   ri   
layer_typegetattrhas_expertsr   s      r4   rq   %GraniteMoeHybridDecoderLayer.__init__a  s    !--##a'$7$?D!6v7I7IvObObc(?@R@RX^XkXk(l%#)#=#= -f5
##I.'93FFDJ6vIDN 229= #6+>BQFr6   rB   rQ   r   output_attentionsr   r   output_router_logitsr   ra   rD   c	                    Un
U R                  U5      nU R                  b  U R                  " SUUUUS.U	D6nSnO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
                  (       a'  U R                  U5      u  pXR                  U5      -   nOU R                  U5      nSnXU R                  -  -   nU4nU(       a  X4-  nU(       a  X4-  nU$ )a  
Args:
    hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
    attention_mask (`torch.FloatTensor`, *optional*):
        attention mask of size `(batch_size, sequence_length)` if flash attention is used or `(batch_size, 1,
        query_sequence_length, key_sequence_length)` if default attention is used.
    past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
    output_attentions (`bool`, *optional*):
        Whether or not to return the attentions tensors of all attention layers. See `attentions` under
        returned tensors for more detail.
    use_cache (`bool`, *optional*):
        If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
        (see `past_key_values`).
    cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
        Indices depicting the position of the input sequence tokens in the sequence
    output_router_logits (`bool`, *optional*):
        Whether or not to return the logits of all the routers. They are useful for computing the router loss, and
        should not be returned during inference.
    position_embeddings (`tuple[torch.FloatTensor, torch.FloatTensor]`, *optional*):
        Tuple containing the cosine and sine positional embeddings of shape `(batch_size, seq_len, head_dim)`,
        with `head_dim` being the embedding dimension of each attention head.
    kwargs (`dict`, *optional*):
        Arbitrary kwargs.Can be used to provide `GraniteFlashAttentionKwargs` for
        padding-free training and/or improve torch.compile performance.
N)rB   r   r)  rQ   )rB   rQ   r   r  r   r   r   r  )r  r   r  r  r  r  r  r  )r   rB   rQ   r   r  r   r   r  r   ra   residualself_attn_weightsmoe_hidden_statesr  outputss                  r4   r   $GraniteMoeHybridDecoderLayer.forwardx  s3   J !,,];::! JJ +-+-	
 M !%/3~~ 	0+--"3#-$7	0 	0,M !43K3K#KK !55mD/3/D/D]/S,-0NNM OOM:M M 43K3K#KK "++G''Gr6   )
r  r  rw   r  r  r   r  r  r  r  )NNFFNFN)ru   r   r   r   r    r   rq   r/   r   r   r   r   r   r   r   r  r   r   r   r   r   s   @r4   r  r  `  s   G5 G# G4 26*.,1$)59/4KOU||U !.U !	U
 $D>U D>U !!1!12U 'tnU &eELL%,,,F&GHU 45U 
u  (51B1BEDUDU1U+V"WW	XU Ur6   r  c                   X   ^  \ rS rSr% \\S'   SrSrS/rS/r	Sr
SrSrSrU 4S jrS	rU =r$ )
GraniteMoeHybridPreTrainedModeli  rk   modelTr  r   Fc                   > [         TU ]  U5        [        U[        5      (       a8  UR                  R
                  R                  SU R                  R                  S9  [        U[        5      (       a  UR                  R
                  R                  S5        [        R                  " [        R                  " SUR                  S-   5      5      UR                   l        UR"                  R
                  R                  S5        g [        U[$        5      (       a&  UR                  R
                  R                  S5        g g )Nr   )r  stdg      ?r   )rp   _init_weights
isinstancer  r:  datanormal_rk   initializer_ranger   r  fill_r/   r  r  ry   r  r#  r!  )r   rM   rt   s     r4   r  -GraniteMoeHybridPreTrainedModel._init_weights  s    f%f=>>MM&&CT[[5R5R&Sf899NN%%c* %		%,,q&:J:JQ:N*O PFLLHHMM$ <==MM$$S) >r6   r  )ru   r   r   r   r    r  base_model_prefixsupports_gradient_checkpointing_no_split_modules_skip_keys_device_placement_supports_flash_attn_supports_sdpa_can_compile_fullgraph_is_statefulr  r   r   r   s   @r4   r  r    sG    ""&*#78#4"5N"L	* 	*r6   r  c                   l   ^  \ rS rSrSS\4U 4S jjjr\R                  " 5       \S 5       5       r	Sr
U =r$ )GraniteMoeHybridRotaryEmbeddingi  rk   c                   > [         TU ]  5         [        US5      (       aZ  [        UR                  [
        5      (       a;  UR                  R                  SUR                  R                  S5      5      U l        OSU l        UR                  U l	        UR                  U l
        Xl        [        U R                     U l        U R                  U R                  U5      u  o0l        U R                  SUSS9  U R                   U l        g )Nrope_scaling	rope_typer}  defaultinv_freqF)
persistent)rp   rq   hasattrr  r  r   getr  max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenrk   r   rope_init_fnattention_scalingregister_bufferr  original_inv_freq)r   rk   r   r  rt   s       r4   rq   (GraniteMoeHybridRotaryEmbedding.__init__  s    6>**z&:M:Mt/T/T#0044[&BUBUBYBYZ`BabDN&DN"("@"@$*$B$B!/?+/+<+<T[[&+Q((ZeD!%r6   c                 b   U R                   S S S 2S 4   R                  5       R                  UR                  S   SS5      R	                  UR
                  5      nUS S 2S S S 24   R                  5       n[        UR
                  R                  [        5      (       a0  UR
                  R                  S:w  a  UR
                  R                  OSn[        R                  " USS9   UR                  5       UR                  5       -  R                  SS5      n[        R                  " Xf4SS	9nUR                  5       U R                  -  nUR                  5       U R                  -  n	S S S 5        WR	                  UR                   S
9W	R	                  UR                   S
94$ ! , (       d  f       N@= f)Nr   r*   r   mpscpuF)device_typeenabledr+   r,   r,  )r  r  rF   r.   r_   r   r  r}  r   r/   autocastr[   r0   r;   r#  r<   rV   )
r   r1   r=   inv_freq_expandedposition_ids_expandedr*  freqsembr;   r<   s
             r4   r   'GraniteMoeHybridRotaryEmbedding.forward  sR    !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.)r#  rk   r   r%  r!  r"  r  r   )ru   r   r   r   r    rq   r/   no_gradr   r   r   r   r   s   @r4   r  r    s7    /5 / /" ]]_<  <r6   r  c                   H  ^  \ rS rSrS\4U 4S jjr\\           SS\R                  S\
\R                     S\
\R                     S\
\\\\R                     4      S\
\R                     S	\
\   S
\
\   S\
\   S\
\   S\
\   S\
\R                     S\\   S\\\4   4S jj5       5       r SS\\R                  S4   S\R                  S\R                  S\S
\4
S jjr\S\R                  S\S\S\R2                  S\R                  S\4S j5       rS rSrU =r$ )GraniteMoeHybridModeli  rk   c           	      6  > [         TU ]  U5        UR                  U l        UR                  U l        [
        R                  " UR                  UR                  U R                  5      U l        [
        R                  " [        UR                  5       Vs/ sH  n[        X5      PM     sn5      U l        [        UR                  UR                  S9U l        SU l        UR$                  U l        UR                  U l        UR&                  U l        U R                  U R(                  -  U l        UR,                  U l        UR.                  U l        UR0                  U l        U R0                  S:X  a  [3        U5      OS U l        U R7                  5         g s  snf )Nr  Frope)rp   rq   pad_token_idpadding_idx
vocab_sizer   	Embeddingrw   embed_tokens
ModuleListr   r   r  layersr  r  r"  gradient_checkpointingembedding_multiplierrx   ry   rK   r  
rope_thetaposition_embedding_typer  
rotary_emb	post_initr   s      r4   rq   GraniteMoeHybridModel.__init__  sE    !.. ++LL):):F<N<NPTP`P`ammNSTZTlTlNmnNm)&<Nmn
 ,F,>,>FDWDWX	&+#$*$?$?!!--33((DNN:'-'E'E$ ++'-'E'E$EIEaEaekEk9&Aqu 	! os   F	input_idsrQ   r=   r   inputs_embedsr   r  output_hidden_statesr  return_dictr   ra   rD   c                    Ub  UOU R                   R                  nUb  UOU R                   R                  nUb  UOU R                   R                  nU
b  U
OU R                   R                  n
US L US L-  (       a  [        S5      eU R                  (       a/  U R                  (       a  U(       a  [        R                  S5        SnUc  U R                  U5      nXPR                  -  nU(       a  Uc  [        R                  S5        UcD  Ub  UR                  5       OSn[        R                  " XUR                  S   -   UR                   S9nUc  UR#                  S5      nU R%                  X%XU5      nU R'                  X+5      nUnS nU R(                  b  U R)                  UU5      nU(       a  SOS nU(       a  SOS nU	(       a  SOS nU R*                   Hj  nUR,                  S	:X  a  UOUnU(       a  UU4-  nU" U4UUUUUU	US
.UD6nUS   nU(       a  US   b	  UUS   4-  nU	(       d  MY  US   c  Ma  UUS   4-  nMl     U R/                  U5      nU(       a  UU4-  nU(       a  UR0                  (       d  SUl        [3        UUUUUS9$ )Nz:You must specify exactly one of input_ids or inputs_embedszX`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.FzGraniteMoeHybrid requires an initialized `HybridMambaAttentionDynamicCache` to return a cache. Because one was not provided, no cache will be returned.r   r   r   r  r   )rQ   r   r  r   r   r  r   r*   T)last_hidden_stater   rB   
attentionsr  )rk   r  rG  r   use_return_dictr|   r>  rX   rr   rs   r;  r?  r   r/   r  r.   r   r8   _update_causal_mask_update_mamba_maskrB  r=  r  r"  r   r   )r   rE  rQ   r=   r   rF  r   r  rG  r  rH  r   ra   past_seen_tokensre   
mamba_maskrB   r   all_hidden_statesall_self_attnsall_router_logitsdecoder_layer
layer_masklayer_outputss                           r4   r   GraniteMoeHybridModel.forward&  s   " 2C1N-TXT_T_TqTq$8$D $++JjJj 	 "+!6IDKK<Q<Q	%0%<k$++B]B]-t";<YZZ&&4==Yj I  --i8M%(A(AA 0K
 !CRC^==?de"\\ ]5H5H5K"KTaThThN )33A6L..>L]
 ,,^L
 &"??&"&//-"N #7BD0d"6BD![[M'4'?'?7'JP[J#!m%55!)
)."3#-%9$7
 
M *!,M  #/"}Q'7&99N## $0%-*;)==%; )> 		-0  -!11?#E#E15O.%+++%+
 	
r6   r'   r   c           	         U R                   R                  S:X  a  Ub  US:H  R                  5       (       a  U$ g U R                   R                  S:X  a,  [        U[        R
                  5      (       a  [        U5      nU$ Ub  UR                  5       OSnUb  UR                  OSnU R                   R                  S:X  a5  U(       d.  U(       d'  [        R                  " UUUU R                  S9(       a  g UR                  nUR                  S   n	U(       a  UR                  5       n
O5[        U[        R
                  5      (       a  UR                  S	   OXi-   S-   n
U R                  UU	U
UUUR                  S   S
9nU R                   R                  S:X  aZ  UbW  UR                   R"                  S;   a=  U(       d6  [        R$                  " U5      R&                  n[        R(                  " X5      nU$ )Nflash_attention_2r   flex_attentionr   Fsdpa)rF  past_key_values_lengthis_trainingr   r*   )sequence_lengthtarget_lengthrV   r   r   )r|  xpunpu)rk   r   anyr  r/   r   r(   r   r   r   _ignore_causal_mask_sdparX   rV   r.   get_max_cache_shape5_prepare_4d_causal_attention_mask_with_cache_positionr   r}  finfomin_unmask_unattended)r   rQ   r   r   r   r  rO  using_compilable_cacherV   r^  r_  re   	min_dtypes                r4   rM  )GraniteMoeHybridModel._update_causal_mask  s    ;;++/BB)~/D.I.I.K.K%%;;++/??.%,,77!<^!L!!
 @O?Z?99;`aCRC^!?!?di ;;++v5>T]n%>>*'7 MM	 ""&,,Q/!+??AM nell;; $$R(%7!;  PP+')#))!, Q 
 KK,,6*%%**.DD%
 E*..I0CCK[Kr6   r^  r_  rV   r   c                    U b  U R                  5       S:X  a  U nU$ [        R                  " U5      R                  n[        R                  " X4XUR
                  S9nUS:w  a  [        R                  " USS9nU[        R                  " X$R
                  S9UR                  SS5      :  -  nUSSSS2SS24   R                  USSS5      nU b  UR                  5       nU R                  S   n	USS2SS2SS2SU	24   U SS2SSSS24   R                  UR
                  5      -   n
U
S:H  n
USS2SS2SS2SU	24   R                  X5      USS2SS2SS2SU	24'   U$ )	a  
Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape
`(batch_size, key_value_length)`, or if the input `attention_mask` is already 4D, do nothing.

Args:
    attention_mask (`torch.Tensor`):
        A 2D attention mask of shape `(batch_size, key_value_length)` or a 4D attention mask of shape
        `(batch_size, 1, query_length, key_value_length)`.
    sequence_length (`int`):
        The sequence length being processed.
    target_length (`int`):
        The target length: when generating with static cache, the mask should be as long as the static cache,
        to account for the 0 padding, the part of the cache that is not filled yet.
    dtype (`torch.dtype`):
        The dtype to use for the 4D attention mask.
    cache_position (`torch.Tensor`):
        Indices depicting the position of the input sequence tokens in the sequence.
    batch_size (`torch.Tensor`):
        Batch size.
Nr   )
fill_valuerV   r   r   r   r   r*   r   )r-   r/   rf  rg  fullr   triur  rG   rF   cloner.   r_   r   )rQ   r^  r_  rV   r   r   ra   re   rj  mask_lengthpadding_masks              r4   re  KGraniteMoeHybridModel._prepare_4d_causal_attention_mask_with_cache_position  s}   < %.*<*<*>!*C(K* ' E*..I** 0Y\j\q\qK !##jjqA5<<>S>STWeWmWmnprsWtttK%dD!Q&67>>z1bRTUK))//1,2226*1aL[L+@ANSTVZ\`bcScDdDgDg&&E    ,q05@Aq,;,AV5W5c5c 6Aq!\k\12 r6   c                 b    UnUS   S:  d!  Ub   [         R                  " US:H  5      (       a  SnU$ )zV
No need for zeroing states when
    1. Cached forward
    2. Attending to all inputs
r   Nr   )r/   all)r   rQ   r   rP  s       r4   rN  (GraniteMoeHybridModel._update_mamba_mask  s:     $
!q ^%?EIIn`aNaDbDbJr6   )r;  r?  r>  rK   rw   r=  r  r"  ry   r8  rA  r@  rB  r9  )NNNNNNNNNNN)F)ru   r   r   r   r    rq   r   r   r/   r   r   r   r   r   listr   r   r   r  r   r   r   rM  staticmethodr   rV   re  rN  r   r   r   s   @r4   r4  r4    s   5 2  '+1537KO59$(,0/3/3&*59s
##s
 !.s
 u//0	s

 "%tE4E4E/F(F"GHs
   1 12s
 D>s
 $D>s
 'tns
 'tns
 d^s
 !!1!12s
 45s
 
u--	.s
  s
v #(BellK78B llB 	B
 B  BH 444 4 {{	4
 4 4 4l	 	r6   r4  gate_logitsr  c                 d   U b  [        U [        5      (       d  g[        U [        5      (       aB  U S   R                  n[        R                  " U  Vs/ sH  oUR                  U5      PM     snSS9n[        R                  R                  R                  WSS9n[        R                  " XrSS9u  p[        R                  R                  R                  X5      n
Uc:  [        R                  " U
R                  5       SS9n[        R                  " USS9nGO"UR                  u  pUR                  S   X-  -  nUSSS2SS2SS4   R                  XXU45      R                  SX!5      R                  W5      n[        R                   " U
R                  5       U-  SS9[        R                   " USS9-  nUSSS2SS2S4   R                  XXR                  S   45      R                  SUR                  S   5      R                  U5      n[        R                   " UU-  SS9[        R                   " USS9-  nUR                  S   [#        UR                  R$                  5      -  n[        R                   " USS2UUUR                  S   -   24   UR'                  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   r,   r*   r   )r  r   r   r/   r0   r_   r   r\   r]   r  one_hotr  r  r.   rF   rG   rV  r   indexr8   )ry  r  r  rQ   compute_device
layer_gateconcatenated_gate_logitsrouting_weightsr   selected_expertsexpert_masktokens_per_expertrouter_prob_per_expertr   r^  r   expert_attention_mask router_per_expert_attention_maskrankoverall_losss                       r4   load_balancing_loss_funcr  $  s   : *[%"@"@+u%%$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&OEZEZ[\E]^_WR..q12R	 	) "'?=]+]cd!ehmhqhq,!i
 "
   #c/*@*@*F*F&GGD99!TD?+@+@+C$CCCDG]GgGghiGjjL +%%a .ks   J-c                      ^  \ rS rSrS/rS\4U 4S jjrS rS r\	             SS\
\R                     S\
\R                     S	\
\R                     S
\
\\\\R"                     4      S\
\R"                     S\
\R                     S\
\   S\
\   S\
\   S\
\   S\
\   S\
\R                     S\\\R                  4   S\\\4   4S jj5       r      SS jrSrU =r$ )GraniteMoeHybridForCausalLMiy  zlm_head.weightrk   c                 J  > [         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                  5         g )NFrn   )rp   rq   r4  r  r9  r   r}   rw   lm_headrouter_aux_loss_coefr  r  r  rC  r  s     r4   rq   $GraniteMoeHybridForCausalLM.__init__|  s     *62
 ++yy!3!3V5F5FUS$*$?$?!!33#)#=#=  	r6   c                     Xl         g r   r  )r   decoders     r4   set_decoder'GraniteMoeHybridForCausalLM.set_decoder  s    
r6   c                     U R                   $ r   r  r   s    r4   get_decoder'GraniteMoeHybridForCausalLM.get_decoder  s    zzr6   rE  rQ   r=   r   rF  labelsr   r  rG  r  rH  r   logits_to_keeprD   c                    Ub  UOU R                   R                  nU
b  U
OU R                   R                  n
U	b  U	OU R                   R                  n	Ub  UOU R                   R                  nU R
                  " SUUUUUUUU	U
UUS.UD6nUS   n[        U[        5      (       a  [        U* S5      OUnU R                  USS2USS24   5      nUU R                   R                  -  nSnUb:  UR                  5       nU R                  " UU4SU R                   R                  0UD6nSnU
(       af  [        U(       a  UR                  OUS   U R                   U R"                  U5      nUb+  UU R$                  UR'                  UR(                  5      -  -  nU(       d!  U4USS -   nU
(       a  U4U-   nUb  U4U-   $ U$ [+        UUUUR,                  UR.                  UR0                  UR                  S9$ )	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, GraniteMoeHybridForCausalLM

>>> model = GraniteMoeHybridForCausalLM.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)rE  rQ   r=   r   rF  r   r  rG  r  rH  r   r   r9  r*   r   )lossaux_lossr  r   rB   rK  r  r  )rk   r  r  rG  rL  r  r  r   slicer  logits_scalingr  loss_functionr9  r  r  r  r  r  r_   r   r   r   rB   rK  )r   rE  rQ   r=   r   rF  r  r   r  rG  r  rH  r   r  ra   r  rB   slice_indicesr  r  r  outputs                         r4   r   #GraniteMoeHybridForCausalLM.forward  s)   P 2C1N-TXT_T_TqTq$8$D $++JjJj 	 %9$D $++JjJj 	 &1%<k$++B]B] ** 
)%+'/!5!5#)
 
   
8B>SV8W8W~ot4]kmA}a,?@A$++444\\^F%%  ;;11 	D /)4%%'"+  ((	H !11HKK4LLLY,F#"v-'+'7D7V#CVC(#33!//))!//
 	
r6   c                 |   US L n	U	(       d]  Uc  US   UR                   S   :  a  US S 2UR                   S   * S 24   nOhUR                   S   UR                   S   :w  a	  US S 2U4   nO>U(       a7  [        U R                  UR                   S   U R                  U R                  S9nUbZ  UcW  UR                  5       R                  S5      S-
  nUR                  US:H  S5        U	(       d  US S 2UR                   S   * S 24   nUb  U	(       a  SU0n
OSUR                  5       0n
U
R                  UUUUUS.5        U
$ )Nr*   r   r   r   rF  rE  )r=   r   r   rQ   r   )
r.   r   rk   rV   r   r  r   masked_fill_r`   r   )r   rE  r   rQ   rF  r   r=   r   ra   empty_past_kvmodel_inputss              r4   prepare_inputs_for_generation9GraniteMoeHybridForCausalLM.prepare_inputs_for_generation  sX    (4/ )!"%);;%a.*>*>q*A)A)C&CD	#~';';A'>>%a&78	>Y__Q/DKKO %,*>)..077;a?L%%n&91= +A	0B/B/D,DE $+];L')=)=)?@L ,#2&"0"0	
 r6   )r  r  r  r  r  r9  )NNNNNNNNNNNNr   )NNNNNT)ru   r   r   r   _tied_weights_keysr    rq   r  r  r   r   r/   r   r   r   r   rw  r   r   r   r   r   r   r  r   r   r   s   @r4   r  r  y  s   *+5   151537KO59-1$(,0/3/3&*5934k
E,,-k
 !.k
 u//0	k

 "%tE4E4E/F(F"GHk
   1 12k
 ))*k
 D>k
 $D>k
 'tnk
 'tnk
 d^k
 !!1!12k
 c5<</0k
  
u//	0!k
 k
` 7 7r6   r  )r  r4  r  )Nr   )r   )Nr+   N)\typingr   r   r   r   r   r/   torch.nn.functionalr   r\   r  transformers.activationsr	   cache_utilsr   r   r   
generationr   modeling_attn_mask_utilsr   modeling_layersr   modeling_outputsr   r   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   r   utils.import_utilsr   r   configuration_granitemoehybridr    +mamba_ssm.ops.triton.selective_state_updater!   !mamba_ssm.ops.triton.ssd_combinedr"   r#   causal_conv1dr$   r%   !torch.nn.attention.flex_attentionr'   integrations.flex_attentionr(   
get_loggerru   rr   r5   rA   r   r   rL   Moduler  rg   ri   r   r   r   r   ru  r%  r   r   r!  r  r  r  r  r  r  r  r  r  r4  r   r  r  __all__r  r6   r4   <module>r     s	  , = <     + < < ) > 9 j j K F & \ \ V B Rmm!DD-7**  !!;J 
		H	%(6	UU\\ 	U# 	U%,, 	U& %II%<<% 
% <<	%
 U\\*% % %:S)		 S)ldou doTVU\\ VS V
(( 46FH\]^ ^_ ^_B;588?? ;$")) 4)5 2Jbii J(*bii *Z-S -S`9+")) 9+xm#= m` *o * *0<bii <D U; U Ut "&
-1	R&u||U5<<%8$>?R&#R& U\\*	R&
 5<<R&j{"A? {| fr6   