+
    "Ý³j¶o  ã                  óª  a € 0 t $ ^ RIHt ^ RIt^ RIt^ RIHt ^ RIHtH	t	H
t
 ^ RIt^ RIHu Ht ^RIHtHt ^RIHtHt ^RIHtHt ^RIHt ^R	IHt  ^ RIt^ R
IHt ^ RI H!t" Rt# ! R R]PL                  4      t' ! R R4      t( ! R R4      t)Rs*R]+R&   R R lt, ! R R]4      t- ! R R]4      t.RR.t/R#   ]$ d   t%RtRtRt"]%t# Rt%A%LfRt%A%ii ; i)é    )ÚannotationsN)ÚPath)ÚDictÚOptionalÚTuple)ÚAdapterÚLora)ÚDEVICEÚPLATFORM)ÚFORMATÚMETHOD)ÚBACKEND)ÚGGUFTorchLinear)Ú	llama_cpp)Úload_shared_libraryc                  óh   € ] tR t^"tR]P
                  3R]P                  3R]P                  3.tRt	R# )Ú_GGMLInitParamsÚmem_sizeÚ
mem_bufferÚno_alloc© N)
Ú__name__Ú
__module__Ú__qualname__Ú__firstlineno__ÚctypesÚc_size_tÚc_void_pÚc_boolÚ_fields_Ú__static_attributes__r   ó    Úf/home/luca/screen-checking/.venv/lib/python3.14/site-packages/gptqmodel/nn_modules/qlinear/gguf_cpp.pyr   r   "   s-   † à	�V—_‘_Ð%Ø	�v—‘Ð'Ø	�V—]‘]Ð#ð„Hr"   r   c                  ó:   € ] tR t^*tR R ltR R ltR R ltRtR# )	Ú_GGMLMatmulPlanc          
     ó,   € V ^8„  d   QhRRRRRRRRRR/# )	é   ÚrowsÚintÚout_featuresÚoutput_nbytesÚoutput_dtypeútorch.dtypeÚreturnÚNoner   )Úformats   "r#   Ú__annotate__Ú_GGMLMatmulPlan.__annotate__+   s>   € ÷ "ñ "ð ð"ð ð"ð ð"ð "ð"ð 
ñ"r"   c               	ó–   € Wn         W n        W0n        W@n        WPn        W`n        Wpn        W€n        W�n        W n	        W°n
        WÀn        R # ©N)ÚctxÚbufferÚweight_tensorÚinput_tensorÚoutput_tensorÚgraphr(   r*   r+   r,   Ú_backend_buffer_freeÚ	_ctx_free)Úselfr5   r6   r7   r8   r9   r:   r(   r*   r+   r,   Úbackend_buffer_freeÚctx_frees   &$$$$$$$$$$$$r#   Ú__init__Ú_GGMLMatmulPlan.__init__+   sG   € ð  ŒØŒØ*ÔØ(ÔØ*ÔØŒ
ØŒ	Ø(ÔØ*ÔØ(ÔØ$7Ô!Ø!Žr"   c               ó   € V ^8„  d   QhRR/# ©r'   r.   r/   r   )r0   s   "r#   r1   r2   H   s   € ÷ ñ �tñ r"   c                	óÚ   € V P                   '       d#   V P                  V P                   4       R V n         V P                  '       d%   V P                  V P                  4       R V n        R # R # r4   )r6   r;   r5   r<   ©r=   s   &r#   ÚcloseÚ_GGMLMatmulPlan.closeH   sI   € Ø�;�;ˆ;Ø×%Ñ% d§k¡kÔ2ØˆDŒKØ�8�8ˆ8Ø�N‰N˜4Ÿ8™8Ô$ØˆDŽHñ r"   c               ó   € V ^8„  d   QhRR/# rC   r   )r0   s   "r#   r1   r2   P   s   € ÷ ñ ˜ñ r"   c                	óL   €  V P                  4        R #   \         d     R # i ; ir4   )rF   Ú	ExceptionrE   s   &r#   Ú__del__Ú_GGMLMatmulPlan.__del__P   s"   € ð	Ø�J‰JŽLøÜô 	âð	ús   ‚ ”#¢#)r;   r<   r6   r5   r:   r8   r*   r,   r+   r9   r(   r7   N)r   r   r   r   r@   rF   rK   r!   r   r"   r#   r%   r%   *   s   † õ"õ:÷ñ r"   r%   c                  ó,  € ] tR t^XtR(tR R ltR R ltR R lt]R R l4       t	R	 R
 lt
R R ltR R ltR R lt]R R l4       t]R R l4       tR R lt]R R l4       t]R R l4       tR R ltR R ltR R  ltR! R" ltR#R$/R% R& lltR'tR$# ))Ú_GGMLBridgec               ó   € V ^8„  d   QhRR/# rC   r   )r0   s   "r#   r1   Ú_GGMLBridge.__annotate__[   s   € ÷ 1ñ 1˜$ñ 1r"   c                	ó"  € \         e   \        R4      \         h\        \        P                  4      P                  4       P                  R,          p\        RV4      V n        \        RV4      V n	        R V n
        R V n         \        RV4      V n
        V P                  4        \        \        P                   4      V n        \        \        P$                  4      V n        R\        \        P(                  4      R\        \        P*                  4      R\        \        P,                  4      R	\        \        P.                  4      R
\        \        P0                  4      /V n        R V n        / V n        R #   \         d   pY n         R p?LöR p?ii ; i)Nz:GGUFCppKernel requires `llama-cpp-python` to be installed.Úlibz	ggml-basezggml-cpuz	ggml-cudaÚQ4_0ÚQ8_0ÚQ4_KÚQ5_KÚQ6_K)Ú_LLAMA_CPP_IMPORT_ERRORÚModuleNotFoundErrorr   Ú_llama_cpp_pkgÚ__file__ÚresolveÚparentÚ_llama_cpp_load_shared_libraryÚ
_ggml_baseÚ	_ggml_cpuÚ
_ggml_cudaÚ_ggml_cuda_errorrJ   Ú_bind_functionsr)   Ú_llama_cpp_libÚGGML_TYPE_F32Úggml_type_f32ÚGGML_TYPE_F16Úggml_type_f16ÚGGML_TYPE_Q4_0ÚGGML_TYPE_Q8_0ÚGGML_TYPE_Q4_KÚGGML_TYPE_Q5_KÚGGML_TYPE_Q6_KÚggml_qtypesÚ_cpu_backendÚ_cuda_backends)r=   Úlib_dirÚexcs   &  r#   r@   Ú_GGMLBridge.__init__[   s:  € Ü"Ò.Ü%ØLóä*ð+ô ”~×.Ñ.Ó/×7Ñ7Ó9×@Ñ@À5ÕHˆÜ8¸ÀgÓNˆŒÜ7¸
ÀGÓLˆŒØˆŒØ59ˆÔð	(Ü<¸[È'ÓRˆDŒOð 	×ÑÔä ¤×!=Ñ!=Ó>ˆÔÜ ¤×!=Ñ!=Ó>ˆÔà”Cœ×5Ñ5Ó6Ø”Cœ×5Ñ5Ó6Ø”Cœ×5Ñ5Ó6Ø”Cœ×5Ñ5Ó6Ø”Cœ×5Ñ5Ó6ð
ˆÔð ,0ˆÔØ.0ˆÖøô ô 	(Ø$'×!Ñ!ûð	(ús   ÂE3 Å3FÅ>F	Æ	Fc               ó    € V ^8„  d   QhRRRR/# )r'   ÚnameÚstrr.   r/   r   )r0   s   "r#   r1   rP   x   s   € ÷ ñ ˜sð ¸$ñ r"   c                	ó4   € \        W4      pW5n        WEn        R # r4   )ÚgetattrÚargtypesÚrestype)r=   rR   ru   ry   rz   Úfns   &&&&& r#   Ú_bindÚ_GGMLBridge._bindx   s   € Ü�SÓˆØŒØŽ
r"   c               ó   € V ^8„  d   QhRR/# rC   r   )r0   s   "r#   r1   rP   }   s   € ÷ $`ñ $` ñ $`r"   c                	óÔ	  € V P                  V P                  R \        .\        P                  4       V P                  V P                  R\        P                  .R4       V P                  V P                  R\        P                  \        P
                  \        P                  \        P                  .\        P                  4       V P                  V P                  R\        P                  \        P                  \        P                  .\        P                  4       V P                  V P                  R\        P                  .R4       V P                  V P                  R\        P                  .\        P                  4       V P                  V P                  R\        P                  \        P                  .R4       V P                  V P                  R\        P                  .\        P                  4       V P                  V P                  R	\        P                  .\        P                  4       V P                  V P                  R
\        P                  \        P                  .\        P                  4       V P                  V P                  R\        P                  .R4       V P                  V P                  R\        P                  .R4       V P                  V P                  R\        P                  \        P                  \        P                  \        P                  .R4       V P                  V P                  R\        P                  \        P                  \        P                  \        P                  .R4       V P                  V P                  R\        P                  \        P                  .\        P
                  4       V P                  V P                  R\        P                  .R4       V P                  V P                  R. \        P                  4       V P                  V P                  R\        P                  \        P
                  .R4       V P                  ej   V P                  V P                  R\        P
                  .\        P                  4       V P                  V P                  R. \        P
                  4       R# R# )Ú	ggml_initÚ	ggml_freeNÚggml_new_tensor_2dÚggml_mul_matÚggml_set_inputÚggml_new_graphÚggml_build_forward_expandÚggml_nbytesÚggml_element_sizeÚggml_backend_alloc_ctx_tensorsÚggml_backend_buffer_freeÚggml_backend_freeÚggml_backend_tensor_setÚggml_backend_tensor_getÚggml_backend_graph_computeÚggml_backend_synchronizeÚggml_backend_cpu_initÚggml_backend_cpu_set_n_threadsÚggml_backend_cuda_initÚ"ggml_backend_cuda_get_device_count)
r|   r_   r   r   r   Úc_intÚc_int64r   r`   ra   rE   s   &r#   rc   Ú_GGMLBridge._bind_functions}   s  € Ø�
‰
�4—?‘? K´/Ð1BÄFÇOÁOÔTØ�
‰
�4—?‘? K´&·/±/Ð1BÀDÔIØ�
‰
Ø�O‰OØ Ü�_‰_œfŸl™l¬F¯N©N¼F¿N¹NÐKÜ�O‰Oô		
ð 	�
‰
�4—?‘? N´V·_±_ÄfÇoÁoÔW]×WfÑWfÐ4gÔio×ixÑixÔyØ�
‰
�4—?‘?Ð$4´v·±Ð6GÈÔNØ�
‰
�4—?‘?Ð$4´v·±Ð6GÌÏÉÔYØ�
‰
�4—?‘?Ð$?Ä&Ç/Á/ÔSY×SbÑSbÐAcÐeiÔjØ�
‰
�4—?‘? M´F·O±OÐ3DÄfÇoÁoÔVØ�
‰
�4—?‘?Ð$7¼&¿/¹/Ð9JÌFÏOÉOÔ\Ø�
‰
�4—?‘?Ð$DÄvÇÁÔX^×XgÑXgÐFhÔjp×jyÑjyÔzØ�
‰
�4—?‘?Ð$>ÄÇÁÐ@QÐSWÔXØ�
‰
�4—?‘?Ð$7¼&¿/¹/Ð9JÈDÔQØ�
‰
Ø�O‰OØ%Ü�_‰_œfŸo™o¬v¯©ÄÇÁÐPØô		
ð 	�
‰
Ø�O‰OØ%Ü�_‰_œfŸo™o¬v¯©ÄÇÁÐPØô		
ð 	�
‰
�4—?‘?Ð$@Ä6Ç?Á?ÔTZ×TcÑTcÐBdÔfl×frÑfrÔsØ�
‰
�4—?‘?Ð$>ÄÇÁÐ@QÐSWÔXØ�
‰
�4—>‘>Ð#:¸BÄÇÁÔPØ�
‰
�4—>‘>Ð#CÄfÇoÁoÔW]×WcÑWcÐEdÐfjÔkØ�?‰?Ò&Ø�J‰J�t—‘Ð(@Ä6Ç<Á<À.ÔRX×RaÑRaÔbØ�J‰J�t—‘Ð(LÈbÔRX×R^ÑR^Ö_ñ 'r"   c               ó   € V ^8„  d   QhRR/# ©r'   r.   r)   r   )r0   s   "r#   r1   rP   ¤   s   € ÷ /ñ /˜#ñ /r"   c                 	óÔ   € \         P                  P                  R 4      p V e    \        ^\	        V 4      4      # \        ^\        P                  ! 4       4      #   \
         d     L,i ; i)ÚGPTQMODEL_GGUF_CPP_THREADS)ÚosÚenvironÚgetÚmaxr)   Ú
ValueErrorÚtorchÚget_num_threads)Úoverrides    r#   Ú_cpu_threadsÚ_GGMLBridge._cpu_threads£   s]   € ä—:‘:—>‘>Ð">Ó?ˆØÒðÜ˜1œc (›mÓ,Ð,ô �1”e×+Ò+Ó-Ó.Ð.øô ô Ùðús   ¥A ÁA'Á&A'c               ó   € V ^8„  d   QhRR/# ©r'   r.   z Tuple[bool, Optional[Exception]]r   )r0   s   "r#   r1   rP   ­   s   € ÷ ñ Ð?ñ r"   c                	ó`   €  V P                  4        R#   \         d   pR T3u Rp?# Rp?ii ; i)FN©TN)Ú_get_cpu_backendrJ   )r=   rr   s   & r#   Úcpu_availableÚ_GGMLBridge.cpu_available­   s7   € ð	Ø×!Ñ!Ô#ð Ðøô ô 	Ø˜#�:Õûð	ús   ‚ ”-Ÿ(¢-¨-c               ó   € V ^8„  d   QhRR/# r¦   r   )r0   s   "r#   r1   rP   ´   s   € ÷ ñ Ð @ñ r"   c                	ó\  €  V P                   f   \        R4      h\        P                  P	                  4       '       g   \        R4      h\        V P                   P                  4       4      pV^ 8:  d   \        R4      hV P                  ^ 4       R#   \         d   pRT3u R p?# R p?ii ; i)Nú5llama-cpp-python was built without GGML CUDA support.zTorch CUDA is unavailable.z(GGML CUDA backend found no CUDA devices.Fr¨   )	ra   ÚRuntimeErrorr    ÚcudaÚis_availabler)   r“   Ú_get_cuda_backendrJ   )r=   Údevice_countrr   s   &  r#   Úcuda_availableÚ_GGMLBridge.cuda_available´   s™   € ð
	Ø�‰Ò&Ü"Ð#ZÓ[Ð[Ü—:‘:×*Ñ*×,Ò,Ü"Ð#?Ó@Ð@Ü˜tŸ™×QÑQÓSÓTˆLØ˜qÔ Ü"Ð#MÓNÐNØ×"Ñ" 1Ô%ð Ðøô ô 	Ø˜#�:Õûð	ús#   ‚<B ¿AB ÂB+ÂB&Â B+Â&B+c               ó   € V ^8„  d   QhRR/# r˜   r   )r0   s   "r#   r1   rP   Â   s   € ÷ !ñ ! #ñ !r"   c                	óî   € V P                   f]   V P                  P                  4       pV'       g   \        R4      hV P                  P	                  WP                  4       4       Wn         V P                   # )Nz4GGUFCppKernel failed to initialize GGML CPU backend.)ro   r`   r�   r¯   r‘   r£   )r=   Úbackends   & r#   r©   Ú_GGMLBridge._get_cpu_backendÂ   s\   € Ø×ÑÒ$Ø—n‘n×:Ñ:Ó<ˆGßÜ"Ð#YÓZÐZØ�N‰N×9Ñ9¸'×CTÑCTÓCVÔWØ 'ÔØ× Ñ Ð r"   c               ó    € V ^8„  d   QhRRRR/# )r'   Údevice_indexr)   r.   r   )r0   s   "r#   r1   rP   Ë   s   € ÷ 1ñ 1¨cð 1°cñ 1r"   c                	ó˜  € V P                   f   \        R4      V P                  hV^ 8  d   ^ p\        V P                   P	                  4       4      pW8¼  d   \        RV RV R24      hWP
                  9  dA   V P                   P                  V4      pV'       g   \        RV R24      hW0P
                  V&   V P
                  V,          # )Nr®   z GGML CUDA backend device index `z` is out of range for `z
` devices.zBGGUFCudaKernel failed to initialize GGML CUDA backend for device `ú`.)ra   r¯   rb   r)   r“   rp   r’   )r=   r»   r³   r¸   s   &&  r#   r²   Ú_GGMLBridge._get_cuda_backendË   sÊ   € Ø�?‰?Ò"ÜÐVÓWÐ]a×]rÑ]rÐrØ˜!ÔØˆLÜ˜4Ÿ?™?×MÑMÓOÓPˆØÔ'ÜØ2°<°.Ð@WÐXdÐWeÐeoÐpóð ð ×2Ñ2Ô2Ø—o‘o×<Ñ<¸\ÓJˆGßÜ"Ð%gÐhtÐguÐuwÐ#xÓyÐyØ07×Ñ Ñ-Ø×"Ñ" <Õ0Ð0r"   c               ó    € V ^8„  d   QhRRRR/# )r'   Úqweightútorch.Tensorr.   r   )r0   s   "r#   r1   rP   Ý   s   € ÷ ñ ¨ð ¸ñ r"   c                	óÌ   € V P                  4       pVP                  P                  R 8w  d   VP                  R R7      pVP	                  4       '       g   VP                  4       pV# )Úcpu©Údevice)ÚdetachrÅ   ÚtypeÚtoÚis_contiguousÚ
contiguous)rÀ   Úqweight_cpus   & r#   Ú_normalize_qweight_cpuÚ"_GGMLBridge._normalize_qweight_cpuÜ   sU   € à—n‘nÓ&ˆØ×Ñ×"Ñ" eÔ+Ø%Ÿ.™.°˜.Ó6ˆKØ×(Ñ(×*Ò*Ø%×0Ñ0Ó2ˆKØÐr"   c               ó$   € V ^8„  d   QhRRRRRR/# )r'   ÚxrÁ   Úpadded_in_featuresr)   r.   r   )r0   s   "r#   r1   rP   æ   s!   € ÷ ñ  ð À#ð È,ñ r"   c                	ó@  € V P                  4       P                  R \        P                  R7      pVP                  R,          V8w  d1   \
        P                  ! V^ WP                  R,          ,
          34      pVP                  4       '       g   VP                  4       pV# )rÃ   ©rÅ   Údtypeéÿÿÿÿ)	rÆ   rÈ   r    Úfloat32ÚshapeÚFÚpadrÉ   rÊ   )rÏ   rÐ   Úx_cpus   && r#   Ú_normalize_input_cpuÚ _GGMLBridge._normalize_input_cpuå   sr   € à—‘“
—‘ U´%·-±-�Ó@ˆØ�;‰;�r�?Ð0Ô0Ü—E’E˜% !Ð%7¿+¹+Àb½/Õ%IÐ!JÓKˆEØ×"Ñ"×$Ò$Ø×$Ñ$Ó&ˆEØˆr"   c               ó$   € V ^8„  d   QhRRRRRR/# )r'   rÏ   rÁ   rÐ   r)   r.   ztuple[torch.Tensor, int]r   )r0   s   "r#   r1   rP   î   s(   € ÷ 'ñ 'àð'ð  ð'ð 
"ñ	'r"   c                	ób  € VP                  4       pVP                  \        P                  8X  d   V P                  pMƒVP                  \        P
                  8X  d   V P                  pMWVP                  \        P                  8X  d.   VP                  \        P                  R 7      pV P                  pM\        R4      hVP                  R,          V8w  d1   \        P                  ! V^ W#P                  R,          ,
          34      pVP                  4       '       g   VP                  4       pW43# )©rÓ   zBGGUFCudaKernel only supports float16, bfloat16, or float32 inputs.rÔ   )rÆ   rÓ   r    Úfloat16rh   rÕ   rf   Úbfloat16rÈ   r¯   rÖ   r×   rØ   rÉ   rÊ   )r=   rÏ   rÐ   Úx_cudaÚggml_input_types   &&&  r#   Ú_normalize_input_cudaÚ!_GGMLBridge._normalize_input_cudaî   sà   € ð
 —‘“ˆØ�<‰<œ5Ÿ=™=Ô(Ø"×0Ñ0‰OØ�\‰\œUŸ]™]Ô*Ø"×0Ñ0‰OØ�\‰\œUŸ^™^Ô+à—Y‘Y¤U§]¡]�YÓ3ˆFØ"×0Ñ0‰OäØTóð ð �<‰<˜ÕÐ1Ô1Ü—U’U˜6 AÐ'9¿L¹LÈÕ<LÕ'LÐ#MÓNˆFØ×#Ñ#×%Ò%Ø×&Ñ&Ó(ˆFØÐ&Ð&r"   c               ó$   € V ^8„  d   QhRRRRRR/# )r'   rÀ   rÁ   rÅ   ztorch.devicer.   r   )r0   s   "r#   r1   rP     s"   € ÷ ñ ¨ð ¸|ð ÐP\ñ r"   c                	ó¸   € V P                  4       pVP                  V8w  d   VP                  VR 7      pVP                  4       '       g   VP	                  4       pV# )rÄ   )rÆ   rÅ   rÈ   rÉ   rÊ   )rÀ   rÅ   Úqweight_cudas   && r#   Ú_normalize_qweight_cudaÚ#_GGMLBridge._normalize_qweight_cuda  sO   € à—~‘~Ó'ˆØ×Ñ &Ô(Ø'Ÿ?™?°&˜?Ó9ˆLØ×)Ñ)×+Ò+Ø'×2Ñ2Ó4ˆLØÐr"   c               ó$   € V ^8„  d   QhRRRRRR/# )r'   Úoutput_element_sizer)   Úkernel_namerv   r.   r-   r   )r0   s   "r#   r1   rP     s#   € ÷ 
ñ 
Àð 
ÐVYð 
Ð^iñ 
r"   c               	ó€   € V ^8X  d   \         P                  # V ^8X  d   \         P                  # \        V RV  R24      h)é   ú0 received unsupported GGML output element size `r½   )r    rÕ   rß   r¯   )rë   rì   s   &$r#   Ú#_torch_dtype_from_ggml_element_sizeÚ/_GGMLBridge._torch_dtype_from_ggml_element_size  sG   € à !Ô#Ü—=‘=Ð Ø !Ô#Ü—=‘=Ð ÜØˆmÐKÐL_ÐK`Ð`bÐcó
ð 	
r"   c               ó8   € V ^8„  d   QhRRRRRRRRRRR	RR
RRR/# )r'   r¸   r)   rË   rÁ   rÙ   Úgguf_tensor_qtyperv   rÐ   r*   rì   r.   r   )r0   s   "r#   r1   rP     sf   € ÷ c+ñ c+ð ðc+ð "ð	c+ð
 ðc+ð ðc+ð  ðc+ð ðc+ð ðc+ð 
ñc+r"   c               	óª  € V P                   P                  \        V P                  R RR7      4      pV'       g   \	        V R24      hR p	 V P                   P                  VV P                  V,          VV4      p
V P                   P                  VV P                  VVP                  ^ ,          4      pV
'       d	   V'       g   \	        V R24      hV P                   P                  V4       V P                   P                  WŠV4      pV'       g   \	        V R24      hV P                   P                  V4      pV'       g   \	        V R24      hV P                   P                  WÜ4       V P                   P                  W�4      p	V	'       g   \	        V R24      hV P                   P                  V
\        P                   ! VP#                  4       4      ^ VP%                  4       VP'                  4       ,          4       V P                   P                  V\        P                   ! VP#                  4       4      ^ VP%                  4       VP'                  4       ,          4       V P                   P)                  W4      pV^ 8w  d   \	        V RV R	24      hV P                   P+                  V4       V P                   P-                  V4      pV P                   P/                  V4      pV^8X  d   \0        P2                  pM)V^8X  d   \0        P4                  pM\	        V R
V R24      h\0        P6                  ! VP                  ^ ,          V3VRR7      pVP%                  4       VP'                  4       ,          pVV8w  d   \	        V RV RV R	24      hV P                   P9                  V\        P                   ! VP#                  4       4      ^ V4       TV	'       d   V P                   P;                  V	4       V P                   P=                  V4       #   T	'       d   T P                   P;                  T	4       T P                   P=                  T4       i ; i)NT©r   r   r   ú, failed to initialize GGML metadata context.ú failed to create GGML tensors.ú# failed to create GGML matmul node.ú failed to allocate GGML graph.ú) failed to allocate GGML backend tensors.ú' GGML graph compute failed with status=Ú.rï   r½   rÃ   ©rÓ   rÅ   ú% GGML output size mismatch: expected ú, got )r_   r€   r   ÚGGML_METADATA_BYTESr¯   r‚   rn   rf   rÖ   r„   rƒ   r…   r†   r‰   rŒ   r   r   Údata_ptrÚnumelÚelement_sizerŽ   r�   r‡   rˆ   r    rÕ   rß   Úemptyr�   rŠ   r�   )r=   r¸   rË   rÙ   ró   rÐ   r*   rì   r5   r6   r7   r8   r9   r:   Ústatusr+   rë   r,   ÚoutputÚexpected_nbytess   &$$$$$$$            r#   Ú_run_quantized_matmulÚ!_GGMLBridge._run_quantized_matmul  s¡  € ð �o‰o×'Ñ'ÜØ×1Ñ1ØØôó
ˆ÷ Ü + Ð.ZÐ[Ó\Ð\àˆðL	+Ø ŸO™O×>Ñ>ØØ× Ñ Ð!2Õ3Ø"Øó	ˆMð  Ÿ?™?×=Ñ=ØØ×"Ñ"Ø"Ø—‘˜A•ó	ˆL÷ !¯Ü" k ]Ð2QÐ#RÓSÐSà�O‰O×*Ñ*¨<Ô8Ø ŸO™O×8Ñ8¸È\ÓZˆMß Ü" k ]Ð2UÐ#VÓWÐWà—O‘O×2Ñ2°3Ó7ˆEßÜ" k ]Ð2QÐ#RÓSÐSØ�O‰O×5Ñ5°eÔKà—_‘_×CÑCÀCÓQˆFßÜ" k ]Ð2[Ð#\Ó]Ð]à�O‰O×3Ñ3ØÜ—’ × 4Ñ 4Ó 6Ó7ØØ×!Ñ!Ó# k×&>Ñ&>Ó&@Õ@ô	ð �O‰O×3Ñ3ØÜ—’ §¡Ó 0Ó1ØØ—‘“ × 2Ñ 2Ó 4Õ4ô	ð —_‘_×?Ñ?ÀÓOˆFØ˜Œ{Ü" k ]Ð2YÐZ`ÐYaÐabÐ#cÓdÐdØ�O‰O×4Ñ4°WÔ=à ŸO™O×7Ñ7¸ÓFˆMØ"&§/¡/×"CÑ"CÀMÓ"RÐØ" aÔ'Ü$Ÿ}™}‘Ø$¨Ô)Ü$Ÿ}™}‘ä"Ø"�mÐ#SÐTgÐShÐhjÐkóð ô —[’[ %§+¡+¨a¥.°,Ð!?À|Ð\aÔbˆFØ$Ÿl™l›n¨v×/BÑ/BÓ/DÕDˆOØ -Ô/Ü"Ø"�mÐ#HÈÐHYÐY_Ð`mÐ_nÐnoÐpóð ð �O‰O×3Ñ3ØÜ—’ §¡Ó 1Ó2ØØô	ð çØ—‘×8Ñ8¸Ô@Ø�O‰O×%Ñ% cÕ*ø÷ Ø—‘×8Ñ8¸Ô@Ø�O‰O×%Ñ% cÕ*ús"   ÁA/P Â;P ÃA=P ÅJP ÐA Qc               ó<   € V ^8„  d   QhRRRRRRRRRRR	RR
RRRRR/	# )r'   r¸   r)   rÀ   rÁ   ró   rv   rÐ   r*   r(   Úinput_ggml_typerì   r.   r%   r   )r0   s   "r#   r1   rP     sq   € ÷ Rñ Rð ðRð ð	Rð
 ðRð  ðRð ðRð ðRð ðRð ðRð 
ñRr"   c               	ó€  € V P                   P                  \        V P                  R RR7      4      p	V	'       g   \	        V R24      hR p
 V P                   P                  V	V P                  V,          VV4      pV P                   P                  V	VVV4      pV'       d	   V'       g   \	        V R24      hV P                   P                  V4       V P                   P                  W›V4      pV'       g   \	        V R24      hV P                   P                  V	4      pV'       g   \	        V R24      hV P                   P                  Wí4       V P                   P                  W‘4      p
V
'       g   \	        V R24      hV P                   P                  V\        P                  ! VP                  4       4      ^ VP!                  4       VP#                  4       ,          4       V P                   P%                  V4      pV P'                  \)        V P                   P+                  V4      4      VR7      p\-        V	V
VVVVVVVVV P                   P.                  V P                   P0                  R	7      #   \2         dA    T
'       d   T P                   P/                  T
4       T P                   P1                  T	4       h i ; i)
NTrõ   rö   r÷   rø   rù   rú   )rì   )r5   r6   r7   r8   r9   r:   r(   r*   r+   r,   r>   r?   )r_   r€   r   r   r¯   r‚   rn   r„   rƒ   r…   r†   r‰   rŒ   r   r   r  r  r  r‡   rð   r)   rˆ   r%   rŠ   r�   rJ   )r=   r¸   rÀ   ró   rÐ   r*   r(   r  rì   r5   r6   r7   r8   r9   r:   r+   r,   s   &$$$$$$$$        r#   Ú build_quantized_matmul_cuda_planÚ,_GGMLBridge.build_quantized_matmul_cuda_plan  si  € ð �o‰o×'Ñ'ÜØ×1Ñ1ØØôó
ˆ÷ Ü + Ð.ZÐ[Ó\Ð\àˆð;	Ø ŸO™O×>Ñ>ØØ× Ñ Ð!2Õ3Ø"Øó	ˆMð  Ÿ?™?×=Ñ=ØØØ"Øó	ˆL÷ !¯Ü" k ]Ð2QÐ#RÓSÐSà�O‰O×*Ñ*¨<Ô8Ø ŸO™O×8Ñ8¸È\ÓZˆMß Ü" k ]Ð2UÐ#VÓWÐWà—O‘O×2Ñ2°3Ó7ˆEßÜ" k ]Ð2QÐ#RÓSÐSØ�O‰O×5Ñ5°eÔKà—_‘_×CÑCÀCÓQˆFßÜ" k ]Ð2[Ð#\Ó]Ð]à�O‰O×3Ñ3ØÜ—’ × 0Ñ 0Ó 2Ó3ØØ—‘“ '×"6Ñ"6Ó"8Õ8ô	ð !ŸO™O×7Ñ7¸ÓFˆMØ×CÑCÜ�D—O‘O×5Ñ5°mÓDÓEØ'ð Dó ˆLô #ØØØ+Ø)Ø+ØØØ)Ø+Ø)Ø$(§O¡O×$LÑ$LØŸ™×2Ñ2ôð øô ô 	ßØ—‘×8Ñ8¸Ô@Ø�O‰O×%Ñ% cÔ*Øð		ús'   ÁAI2 Â I2 Â(A=I2 Ä&EI2 É2J=Ê8J=c          
     ó,   € V ^8„  d   QhRRRRRRRRR	R/# )
r'   r¸   r)   Úplanr%   rÏ   rÁ   rì   rv   r.   r   )r0   s   "r#   r1   rP   Ó  s>   € ÷ ñ ð ðð ð	ð
 ðð ðð 
ñr"   c               	ód  € V P                   P                  VP                  \        P                  ! VP                  4       4      ^ VP                  4       VP                  4       ,          4       V P                   P                  WP                  4      pV^ 8w  d   \        V RV R24      hV P                   P                  V4       \        P                  ! VP                  VP                  3VP                   VP"                  R7      pVP                  4       VP                  4       ,          pWrP$                  8w  d   \        V RV RVP$                   R24      hV P                   P'                  VP(                  \        P                  ! VP                  4       4      ^ VP$                  4       V# )r   rû   rü   rý   rþ   rÿ   )r_   rŒ   r8   r   r   r  r  r  rŽ   r:   r¯   r�   r    r  r(   r*   r,   rÅ   r+   r�   r9   )r=   r¸   r  rÏ   rì   r  r  r  s   &$$$$   r#   Úrun_quantized_matmul_cuda_planÚ*_GGMLBridge.run_quantized_matmul_cuda_planÓ  s]  € ð 	�‰×/Ñ/Ø×ÑÜ�OŠO˜AŸJ™J›LÓ)ØØ�G‰G‹I˜Ÿ™Ó(Õ(ô		
ð —‘×;Ñ;¸GÇZÁZÓPˆØ�QŒ;Ü + Ð.UÐV\ÐU]Ð]^Ð_Ó`Ð`Ø�‰×0Ñ0°Ô9ä—’˜dŸi™i¨×):Ñ):Ð;À4×CTÑCTÐ]^×]eÑ]eÔfˆØ Ÿ,™,›.¨6×+>Ñ+>Ó+@Õ@ˆØ×0Ñ0Ô0ÜØ�-ÐDÀ_ÐDUÐU[Ð\`×\nÑ\nÐ[oÐopÐqóð ð 	�‰×/Ñ/Ø×ÑÜ�OŠO˜FŸO™OÓ-Ó.ØØ×Ñô		
ð ˆr"   c               ó0   € V ^8„  d   QhRRRRRRRRRRR	R/# )
r'   rÀ   rÁ   rÏ   ró   rv   rÐ   r)   r*   r.   r   )r0   s   "r#   r1   rP   ô  sH   € ÷ 
ñ 
ð ð
ð ð	
ð
 ð
ð  ð
ð ð
ð 
ñ
r"   c          
     	óÖ   € VP                   P                  R 8w  d   \        R4      hV P                  V P	                  4       V P                  V4      V P                  W$4      VVVRR7      # )rÃ   z.GGUFCppKernel only supports CPU input tensors.ÚGGUFCppKernel)r¸   rË   rÙ   ró   rÐ   r*   rì   )rÅ   rÇ   r¯   r  r©   rÌ   rÚ   )r=   rÀ   rÏ   ró   rÐ   r*   s   &$$$$$r#   Úquantized_matmul_cpuÚ _GGMLBridge.quantized_matmul_cpuô  sl   € ð �8‰8�=‰=˜EÔ!ÜÐOÓPÐPà×)Ñ)Ø×)Ñ)Ó+Ø×3Ñ3°GÓ<Ø×+Ñ+¨AÓBØ/Ø1Ø%Ø'ð *ó 
ð 	
r"   r  Nc               ó4   € V ^8„  d   QhRRRRRRRRRRR	R
RR/# )r'   rÀ   rÁ   rÏ   ró   rv   rÐ   r)   r*   r  z_GGMLMatmulPlan | Noner.   z$tuple[torch.Tensor, _GGMLMatmulPlan]r   )r0   s   "r#   r1   rP   
  sR   € ÷ "ñ "ð ð"ð ð	"ð
 ð"ð  ð"ð ð"ð %ð"ð 
.ñ"r"   c               	ó’  € VP                   P                  R 8w  d   \        R4      hVP                   pT P                  VP                  f   ^ MVP                  4      pV P                  W$4      w  ršVf=   V P                  WR7      pV P                  VVVVVV	P                  ^ ,          V
RR7      pV P                  VVV	RR7      pWÆ3# )r°   z0GGUFCudaKernel only supports CUDA input tensors.rÄ   ÚGGUFCudaKernel)r¸   rÀ   ró   rÐ   r*   r(   r  rì   )r¸   r  rÏ   rì   )
rÅ   rÇ   r¯   r²   Úindexrã   rè   r  rÖ   r  )r=   rÀ   rÏ   ró   rÐ   r*   r  rÅ   r¸   rá   r  rç   r  s   &$$$$$$      r#   Úquantized_matmul_cudaÚ!_GGMLBridge.quantized_matmul_cuda
  sÕ   € ð �8‰8�=‰=˜FÔ"ÜÐQÓRÐRà—‘ˆØ×(Ñ(¨f¯l©lÒ.B©ÈÏÉÓUˆØ"&×"<Ñ"<¸QÓ"SÑˆØŠ<Ø×7Ñ7¸Ð7ÓOˆLØ×8Ñ8ØØ$Ø"3Ø#5Ø)Ø—\‘\ !•_Ø /Ø,ð 9ó 	ˆDð ×4Ñ4ØØØØ(ð	 5ó 
ˆð ˆ|Ðr"   )	ro   rp   r_   r`   ra   rb   rn   rh   rf   i   )r   r   r   r   r   r@   r|   rc   Ústaticmethodr£   rª   r´   r©   r²   rÌ   rÚ   rã   rè   rð   r  r  r  r  r  r!   r   r"   r#   rN   rN   X   sÃ   † Ø!Ðõ1õ:õ
$`ðL ô/ó ð/õõõ!õ1ð" ôó ðð ôó ðõ'ð2 ôó ðð ô
ó ð
õc+õJRõhõB
ð,"ð (,÷"ó "r"   rN   zOptional[_GGMLBridge]Ú_GGML_BRIDGEc               ó   € V ^8„  d   QhRR/# )r'   r.   rN   r   )r0   s   "r#   r1   r1   2  s   € ÷ ñ ˜+ñ r"   c                 ó2   € \         f   \        4       s \         # r4   )r   rN   r   r"   r#   Ú_get_ggml_bridger#  2  s   € äÒÜ"“}ˆÜÐr"   c                  óÂ  a € ] tR tRt]P
                  .t]P                  .t	]
P                  ^/t. ROtRtRtRt^.t^.t]P&                  .t]P,                  .t]P2                  ]P4                  ]P6                  .t].t]P>                  ]P@                  ]PB                  .t"Rt#Rt$Rt%Rt&Rt'Rt(Rt)R]P6                  RR3R V 3R lllt*]+R R	 l4       t,R
 R lt-Rt.V ;t/# )r  i9  TFÚggufNc               ó<   € V ^8„  d   QhRRRRRRRRRRRRR	R
RRRR/	# ©r'   Ú
group_sizer)   ÚsymÚboolÚdesc_actÚin_featuresr*   ÚbiasÚ
pack_dtyper-   Úadapterr   Úregister_buffersr   )r0   s   "r#   r1   ÚGGUFCppKernel.__annotate__S  sf   € ÷ 
ñ 
ð ð
ð ð	
ð
 ð
ð ð
ð ð
ð ð
ð  ð
ð ð
ð ñ
r"   c                	ó�   <€ VP                  R \        P                  4       \        SV `  ! RRVRVRVRVRVRVRVRVR	V	R
V
/
VB  R# ©r¸   Úbitsr(  r)  r+  r,  r*   r-  r.  r/  r0  Nr   )Ú
setdefaultr   ÚGGUF_CPP_CPUÚsuperr@   ©r=   r4  r(  r)  r+  r,  r*   r-  r.  r/  r0  ÚkwargsÚ	__class__s   &&&&&&&&&&&,€r#   r@   ÚGGUFCppKernel.__init__S  s’   ø€ ð 	×Ñ˜)¤W×%9Ñ%9Ô:Ü‰Òñ 	
Øð	
à!ð	
ð ð	
ð ð		
ð
 $ð	
ð &ð	
ð ð	
ð "ð	
ð ð	
ð .ð	
ð ô	
r"   c               ó   € V ^8„  d   QhRR/# r¦   r   )r0   s   "r#   r1   r1  q  ó   € ÷ ñ Ð>ñ r"   c                	ól   €  \        4       P                  4       #   \         d   pR T3u Rp?# Rp?ii ; i©FN)r#  rª   rJ   ©Úclsrr   s   & r#   Úvalidate_onceÚGGUFCppKernel.validate_oncep  s3   € ð	Ü#Ó%×3Ñ3Ó5Ð5øÜô 	Ø˜#�:Õûð	úó   ‚ š3¥.¨3®3c               ó   € V ^8„  d   QhRR/# ©r'   rÏ   rÁ   r   )r0   s   "r#   r1   r1  w  s   € ÷ .ñ .˜ñ .r"   c                	óz  € VP                   R R V P                  3,           pVP                  RVP                   R,          4      pVP                  P                  R8w  d#   \        V P                  P                   R24      h\        4       P                  V P                  VV P                  V P                  V P                  R7      pVP                  VP                  8w  d   VP                  VP                  R7      pV P                  er   V P                  pVP                  VP                  8w  g   VP                  VP                  8w  d(   VP                  VP                  VP                  R7      pWE,           pV P                   '       d   V P                   P#                  W4R7      pVP                  V4      # )NrÃ   z| only supports CPU inference. Load GGUF models on CPU or use BACKEND.GGUF_CPP_CUDA or BACKEND.GGUF_TORCH for CUDA inference.)rÀ   rÏ   ró   rÐ   r*   rÞ   rÒ   ©rÏ   ÚoutrÔ   )rÖ   r*   ÚreshaperÅ   rÇ   r¯   r:  r   r#  r  rÀ   ró   rÐ   rÓ   rÈ   r-  r/  Úapply)r=   rÏ   Úoriginal_shapeÚx_flatr  r-  s   &&    r#   ÚforwardÚGGUFCppKernel.forwardw  sY  € ØŸ™  "˜¨×):Ñ):Ð(<Õ<ˆØ—‘˜2˜qŸw™w r�{Ó+ˆØ�=‰=×Ñ Ô&ÜØ—>‘>×*Ñ*Ð+ð ,qð qóð ô
 "Ó#×8Ñ8Ø—L‘LØØ"×4Ñ4Ø#×6Ñ6Ø×*Ñ*ð 9ó 
ˆð �<‰<˜6Ÿ<™<Ô'Ø—Y‘Y V§\¡\�YÓ2ˆFà�9‰9Ò Ø—9‘9ˆDØ�{‰{˜fŸm™mÔ+¨t¯z©z¸V¿\¹\Ô/IØ—w‘w f§m¡m¸6¿<¹<�wÓH�Ø•]ˆFà�<�<ˆ<Ø—\‘\×'Ñ'¨&Ð'Ó=ˆFà�~‰~˜nÓ-Ð-r"   r   ©rî   é   é   é   )0r   r   r   r   r   r6  ÚSUPPORTS_BACKENDSr   ÚGGUFÚSUPPORTS_METHODSr   ÚSUPPORTS_FORMATSÚSUPPORTS_BITSÚSUPPORTS_SHARDSÚSUPPORTS_TRAININGÚSUPPORTS_AUTO_PADDINGÚ!SUPPORTS_IN_FEATURES_DIVISIBLE_BYÚ"SUPPORTS_OUT_FEATURES_DIVISIBLE_BYr
   ÚCPUÚSUPPORTS_DEVICESr   ÚALLÚSUPPORTS_PLATFORMr    Úint8Úint16Úint32ÚSUPPORTS_PACK_DTYPESr	   ÚSUPPORTS_ADAPTERSrß   rà   rÕ   ÚSUPPORTS_DTYPESÚREQUIRES_FORMAT_V2ÚAUTOTUNEÚ
QUANT_TYPEÚpackÚ
pack_blockÚpack_gpuÚpack_originalr@   ÚclassmethodrB  rN  r!   Ú__classcell__©r:  s   @r#   r  r  9  s  ø† Ø ×-Ñ-Ð.ÐØŸ™�}ÐØŸ™ RÐ(ÐÚ €MØ€OØÐØ ÐØ)*¨Ð%Ø*+¨Ð&ØŸ
™
�|ÐØ!Ÿ™˜ÐØ!ŸJ™J¨¯©°U·[±[ÐAÐØ˜ÐØ—}‘} e§n¡n°e·m±mÐD€OàÐØ€Hà€Jà€DØ€JØ€HØ€Mð Ø"'§+¡+ØØ!%÷
ò 
ð: ôó ð÷.ó .r"   r  c                  óà  a € ] tR tRt]P
                  .t]P                  .t	]
P                  ^#/t. ROtRtRtRt^.t^.t]P&                  .t]P,                  .t]P2                  ]P4                  ]P6                  .t].t]P>                  ]P@                  ]PB                  .t"Rt#Rt$Rt%Rt&Rt'Rt(Rt)R]P6                  RR3R V 3R lllt*]+R R	 l4       t,R
 V 3R llt-R R lt.R R lt/Rt0V ;t1# )r  i•  TFr%  Nc               ó<   € V ^8„  d   QhRRRRRRRRRRRRR	R
RRRR/	# r'  r   )r0   s   "r#   r1   ÚGGUFCudaKernel.__annotate__¯  sq   € ÷ ]ñ ]ð ð]ð ð	]ð
 ð]ð ð]ð ð]ð ð]ð  ð]ð ð]ð ñ]r"   c                	óž   <€ VP                  R \        P                  4       \        SV `  ! RRVRVRVRVRVRVRVRVR	V	R
V
/
VB  / V n        R# r3  )r5  r   ÚGGUF_CPP_CUDAr7  r@   Ú_ggml_cuda_plansr8  s   &&&&&&&&&&&,€r#   r@   ÚGGUFCudaKernel.__init__¯  sž   ø€ ð 	×Ñ˜)¤W×%:Ñ%:Ô;Ü‰Òñ 	
Øð	
à!ð	
ð ð	
ð ð		
ð
 $ð	
ð &ð	
ð ð	
ð "ð	
ð ð	
ð .ð	
ð ò	
ð []ˆÖr"   c               ó   € V ^8„  d   QhRR/# r¦   r   )r0   s   "r#   r1   rt  Î  r=  r"   c                	ól   €  \        4       P                  4       #   \         d   pR T3u Rp?# Rp?ii ; ir?  )r#  r´   rJ   r@  s   & r#   rB  ÚGGUFCudaKernel.validate_onceÍ  s3   € ð	Ü#Ó%×4Ñ4Ó6Ð6øÜô 	Ø˜#�:Õûð	úrD  c               ó   € V ^8„  d   QhRR/# rC   r   )r0   s   "r#   r1   rt  Ô  s   € ÷ ,ñ , Dñ ,r"   c                	ó¶   <€ V P                   P                  4        F  pVP                  4        K  	  V P                   P                  4        \        SV `  4       # r4   )rw  ÚvaluesrF   Úclearr7  Úclear_weight_cache)r=   r  r:  s   & €r#   r€  Ú!GGUFCudaKernel.clear_weight_cacheÔ  sE   ø€ Ø×)Ñ)×0Ñ0Ö2ˆDØ�J‰JŽLñ 3à×Ñ×#Ñ#Ô%Ü‰wÑ)Ó+Ð+r"   c               ó    € V ^8„  d   QhRRRR/# )r'   rM  rÁ   r.   z!tuple[int, int, torch.dtype, int]r   )r0   s   "r#   r1   rt  Ú  s   € ÷ 
ñ 
 \ð 
Ð6Wñ 
r"   c                	óÔ   € VP                   P                  f   ^ MVP                   P                  pVVP                  ^ ,          VP                  V P                  P                  4       3# r4   )rÅ   r  rÖ   rÓ   rÀ   r  )r=   rM  r»   s   && r#   Ú_cuda_plan_keyÚGGUFCudaKernel._cuda_plan_keyÚ  sQ   € Ø"ŸM™M×/Ñ/Ò7‘q¸V¿]¹]×=PÑ=PˆàØ�L‰L˜�OØ�L‰LØ�L‰L×!Ñ!Ó#ð	
ð 	
r"   c               ó   € V ^8„  d   QhRR/# rF  r   )r0   s   "r#   r1   rt  ã  s   € ÷ .ñ .˜ñ .r"   c           
     	óV  € VP                   R R V P                  3,           pVP                  RVP                   R,          4      pVP                  P                  R8w  d#   \        V P                  P                   R24      hV P                  V4      p\        4       P                  V P                  VV P                  V P                  V P                  V P                  P                  V4      R7      w  rVV P                  P!                  WF4       VP                  VP                  8w  g   VP"                  VP"                  8w  d(   VP%                  VP                  VP"                  R7      pV P&                  er   V P&                  pVP                  VP                  8w  g   VP"                  VP"                  8w  d(   VP%                  VP                  VP"                  R7      pWW,           pV P(                  '       d   V P(                  P+                  W5R7      pVP                  V4      # )Nr°   z| only supports CUDA inference. Load GGUF models on CUDA or use BACKEND.GGUF_CPP_CPU or BACKEND.GGUF_TORCH for CPU inference.)rÀ   rÏ   ró   rÐ   r*   r  rÒ   rH  rÔ   )rÖ   r*   rJ  rÅ   rÇ   r¯   r:  r   r„  r#  r  rÀ   ró   rÐ   rw  r�   r5  rÓ   rÈ   r-  r/  rK  )r=   rÏ   rL  rM  Úplan_keyr  r  r-  s   &&      r#   rN  ÚGGUFCudaKernel.forwardã  s©  € ØŸ™  "˜¨×):Ñ):Ð(<Õ<ˆØ—‘˜2˜qŸw™w r�{Ó+ˆØ�=‰=×Ñ Ô'ÜØ—>‘>×*Ñ*Ð+ð ,pð póð ð
 ×&Ñ& vÓ.ˆÜ'Ó)×?Ñ?Ø—L‘LØØ"×4Ñ4Ø#×6Ñ6Ø×*Ñ*Ø×&Ñ&×*Ñ*¨8Ó4ð @ó 
‰ˆð 	×Ñ×(Ñ(¨Ô8Ø�=‰=˜FŸM™MÔ)¨V¯\©\¸V¿\¹\Ô-IØ—Y‘Y f§m¡m¸6¿<¹<�YÓHˆFà�9‰9Ò Ø—9‘9ˆDØ�{‰{˜fŸm™mÔ+¨t¯z©z¸V¿\¹\Ô/IØ—w‘w f§m¡m¸6¿<¹<�wÓH�Ø•]ˆFà�<�<ˆ<Ø—\‘\×'Ñ'¨&Ð'Ó=ˆFà�~‰~˜nÓ-Ð-r"   )rw  rP  )2r   r   r   r   r   rv  rT  r   rU  rV  r   rW  rX  rY  rZ  r[  r\  r]  r
   ÚCUDAr_  r   r`  ra  r    rb  rc  rd  re  r	   rf  rß   rà   rÕ   rg  rh  ri  rj  rk  rl  rm  rn  r@   ro  rB  r€  r„  rN  r!   rp  rq  s   @r#   r  r  •  s  ø† Ø ×.Ñ.Ð/ÐØŸ™�}ÐØŸ™ RÐ(ÐÚ €MØ€OØÐØ ÐØ)*¨Ð%Ø*+¨Ð&ØŸ™�}ÐØ!Ÿ™˜ÐØ!ŸJ™J¨¯©°U·[±[ÐAÐØ˜ÐØ—}‘} e§n¡n°e·m±mÐD€OàÐØ€Hà€Jà€DØ€JØ€HØ€Mð Ø"'§+¡+ØØ!%÷]ò ]ð< ôó ð÷,ð ,õ
÷.ó .r"   r  )0Ú__conditional_annotations__Ú
__future__r   r   r›   Úpathlibr   Útypingr   r   r   r    Útorch.nn.functionalÚnnÚ
functionalr×   Úadapter.adapterr   r	   Úmodels._constr
   r   Úquantization.configr   r   Úutils.backendr   r%  r   r   rZ   rd   Úllama_cpp._ctypes_extensionsr   r^   rX   rJ   rr   Ú	Structurer   r%   rN   r   Ú__annotations__r#  r  r  Ú__all__)r‹  s   @r#   Ú<module>rš     sÙ   øð÷ #Ð "ã Û 	Ý ß (Ñ (ã ß Ð ç ,ß -ß 1Ý $Ý !ð
"Û&Ý5Ýbà"Ðô�f×&Ñ&ô ÷+ñ +÷\Tñ Tðn '+€Ð#Ó *õôZ.�Oô Z.ôxm.�_ô m.ð` Ð,Ð
-‚øðU ô "Ø€NØ€NØ%)Ð"Ø!Õûð	"ús   ÁB9 Â9CÃ CÃC