+
    "Ý³jâ2  ã                  ó  a € 0 t $ ^ RIHt ^ RIHtHt ^ RIHtHtH	t	H
t
 ^ RItRtR]R&   RtR]R&   R	tRtRtRt ]! R
4      t ^ RIHt ^ RIHt  ^ RIHt ]t]t]tRtR R lt]! 4       tR R ltR R ltR R ltR t R t!R t"R R lt#R R lt$R R lt%R  R! lt&R" t']PP                  ! 4       R# 4       t)R#   ] d     L�i ; i  ] d     ^ RIHt  L�  ] d    Rt  L�i ; ii ; i  ] d   t]t RtAL›RtAii ; i)$é    )Úannotations)ÚPackageNotFoundErrorÚversion)ÚAnyÚDictÚMappingÚOptionalNzOptional[str]ÚVLLM_VERSIONzOptional[Exception]ÚVLLM_IMPORT_ERRORFÚvllm)ÚLLM)ÚSamplingParams)ÚTokensPromptTc               ó   € V ^8„  d   QhRR/# )é   ÚreturnÚstr© )Úformats   "ÚU/home/luca/screen-checking/.venv/lib/python3.14/site-packages/gptqmodel/utils/vllm.pyÚ__annotate__r   /   s   € ÷ 	Nñ 	N 3ñ 	Nó    c                 óÈ   € \         f   \        \        \        4      '       d   R# \        e6   R\         ;'       g    R R\	        \        4      P
                   R\         R2# R# )Nz@vLLM is not installed. Please install via `pip install -U vllm`.úvLLM zwith unknown versionz$ is installed but failed to import: z: zc. Check that vLLM's Python dependencies and its PyTorch/CUDA runtime build are mutually compatible.)r
   Ú
isinstancer   ÚModuleNotFoundErrorÚtypeÚ__name__r   r   r   Ú_vllm_unavailable_messager   /   se   € ÜÒ¤
Ô+<Ô>Q× RÒ RÙQÜÒ$à”L×:Ð:Ð$:Ð;Ð;_ÜÔ%Ó&×/Ñ/Ð0°Ô3DÐ2Eð Fpðpð	
ñ
 Nr   c               ó   € V ^8„  d   QhRR/# )r   r   ÚNoner   )r   s   "r   r   r   >   s   € ÷ Cñ C�tñ Cr   c                 óF   € \         '       g   \        \        4      \        hR # ©N)ÚVLLM_AVAILABLEÚ
ValueErrorÚVLLM_INSTALL_HINTr   r   r   r   Ú_require_vllmr'   >   s   € ß‹>ÜÔ*Ó+Ô1BÐBñ r   c               ó    € V ^8„  d   QhRRRR/# )r   Úvaluer   r   z	list[int]r   )r   s   "r   r   r   C   s   € ÷ 	Tñ 	T Cð 	T¨Iñ 	Tr   c                óÈ  € \         P                  ! V 4      '       d-   V P                  4       P                  4       P	                  4       p \        V \        4      '       d   \        V \        4      '       g   V .# \        V \        \        34      '       dF   \        ;QJ d    R  V  4       F  '       d   K   RM	  RM! R  V  4       4      '       d   \        V 4      # \        R4      h)c              3  ó|   "  € T F2  p\        V\        4      ;'       d    \        V\        4      '       * x € K4  	  R # 5ir#   ©r   ÚintÚbool©Ú.0Úitems   & r   Ú	<genexpr>Ú+_normalize_eos_token_ids.<locals>.<genexpr>H   s/   é € ð 0ÙINÀŒ
�4œÓ×<Ð<¤j°´tÓ&<Ô"<Ô<Ëùó   ‚<ž<FTz<`eos_token_id` must be an integer or a sequence of integers.)ÚtorchÚ	is_tensorÚdetachÚcpuÚtolistr   r-   r.   ÚlistÚtupleÚallÚ	TypeError)r)   s   &r   Ú_normalize_eos_token_idsr>   C   s¥   € Ü‡‚�u×ÒØ—‘“×"Ñ"Ó$×+Ñ+Ó-ˆÜ�%œ×Ò¤j°¼×&=Ò&=ØˆwˆÜ�%œ$¤˜×'Ò'¯C«Cñ 0ÙINó0¯C¯CªCñ 0ÙINó0÷ -ò -ô �E‹{ÐÜ
ÐRÓ
SÐSr   c               ó   € V ^8„  d   QhRR/# )r   Ú	hf_paramszDict[str, Any]r   )r   s   "r   r   r   P   s   € ÷ 8ñ 8¨ñ 8r   c                óð  € \        4        V P                  R 4      e   \        R4      hV P                  R4      e   \        R4      hV P                  R4      R"9  d   \        R4      hV P                  R4      RJ d,   V P                  R	4      ;'       g    ^^8”  d   \        R
4      h/ R	RbRRbRRbRRbRRbRRbRRbRRbRRbRRbRRbRRbRRbRRbRRbRRbpVP                  4        UUu/ uF#  w  r#V P                  V4      f   K  W0V,          bK%  	  pppV P                  R4      eS   \	        V R,          4      pVP                  R4      ;'       g    . p\        \        P                  . VOVO4      4      VR&   V P                  R4      RJ d   RVR&    \        R#/ VB # u uppi   \         dA   pRP                  \        T4      4      p\        R\        ;'       g    R R T R!24      ThRp?ii ; i)$Ú
max_lengthNzQvLLM does not support argument `max_length`. Please use `max_new_tokens` instead.Ú
min_lengthzQvLLM does not support argument `min_length`. Please use `min_new_tokens` instead.Ú	num_beamsztGPT-QModel's vLLM generation adapter does not support Hugging Face beam search. Use vLLM's beam-search API directly.Ú	do_sampleFÚnum_return_sequenceszzvLLM requires sampling when `num_return_sequences` is greater than one. Set `do_sample=True` or request a single sequence.ÚnÚrepetition_penaltyÚtemperatureÚtop_kÚtop_pÚmin_pÚmax_new_tokensÚ
max_tokensÚmin_new_tokensÚ
min_tokensÚfrequency_penaltyÚpresence_penaltyÚ
ignore_eosÚstopÚstop_token_idsÚseedÚeos_token_idg        z, r   Úunknownz5 does not support the requested sampling parameters: Ú.)Né   r   )r'   Úgetr%   Úitemsr>   r:   ÚdictÚfromkeysr   r=   ÚjoinÚsortedr
   )	r@   Ú	field_mapÚsourceÚtargetÚsampling_kwargsÚeos_token_idsÚexisting_stop_idsÚexcÚ	requesteds	   &        r   Úconvert_hf_params_to_vllmri   P   s’  € Ü„Oà‡}�}�\Ó"Ò.ÜÐlÓmÐmØ‡}�}�\Ó"Ò.ÜÐlÓmÐmØ‡}�}�[Ó!¨Ô2Üð3ó
ð 	
ð ‡}�}�[Ó! UÓ*°	·±Ð>TÓ0U×0ZÐ0ZÐYZÐ^_Ô/_ÜðAó
ð 	
ð
Ø ðàÐ2ðð 	�}ðð 	�ð	ð
 	�ðð 	�ðð 	˜,ðð 	�lðð 	˜,ðð 	�lðð 	Ð0ðð 	Ð.ðð 	�lðð 	�ðð 	Ð*ðð  	�ð!€Ið( (Ÿo™oÔ/ôá/‰NˆFØ�=‰=˜Ó ô 	"ˆ˜&Õ!Ò!Ù/ð ñ ð
 ‡}�}�^Ó$Ò0Ü0°¸>Õ1JÓKˆØ+×/Ñ/Ð0@ÓA×GÐGÀRÐÜ,0´·±Ð?cÐARÐ?cÐUbÐ?cÓ1dÓ,eˆÐ(Ñ)Ø‡}�}�[Ó! UÓ*ð *-ˆ˜Ñ&ðÜÑ0 Ñ0Ð0ùóøô  ô Ø—I‘Iœf _Ó5Ó6ˆ	ÜØ”L×-Ð- IÐ.Ð.cÐdmÐcnÐnoÐpó
àð	ûðús$   Ã-F$Ä	F$Æ
F* Æ*G5Æ5;G0Ç0G5c                ó0   € \        4        \        RR V /VB # )Úmodelr   )r'   r   )rk   Úkwargss   &,r   Úload_model_by_vllmrm   ‹   s&   € ô „OÜñ Øðà
ñð r   c                ó  € \        V R R4      pVe   V# \        V RR4      pVf   \        R4      h\        VRR4      ;'       g    \        V RR4      p\        VR R4      pVf   \        VR R4      pVf   \        R4      hV# )Úmodel_configNÚ
llm_enginez*vLLM LLM instance is missing `llm_engine`.Úvllm_configzJvLLM engine exposes neither `vllm_config.model_config` nor `model_config`.)ÚgetattrÚAttributeError)rk   ro   Úenginerq   s   &   r   Úget_vllm_model_configru   –   s•   € Ü˜5 .°$Ó7€LØÒØÐä�U˜L¨$Ó/€FØ‚~ÜÐIÓJÐJä˜& -°Ó6×]Ð]¼'À%ÈÐX\Ó:]€KÜ˜;¨¸Ó=€LØÒÜ˜v ~°tÓ<ˆØÒÜÐiÓjÐjØÐr   c                óÂ   € \        V R R4      pVf   R# \        VRR4      ;'       g    \        V RR4      p\        VRR4      pVf   \        VRR4      p\        VRR4      # )rp   Nrq   Údevice_configÚdevice)rr   )rk   rt   rq   rw   s   &   r   Úget_vllm_devicery   ¨   si   € Ü�U˜L¨$Ó/€FØ‚~Ùä˜& -°Ó6×]Ð]¼'À%ÈÐX\Ó:]€KÜ˜K¨¸$Ó?€MØÒÜ ¨¸Ó>ˆÜ�= (¨DÓ1Ð1r   c               ó    € V ^8„  d   QhRRRR/# )r   r)   r   r   úlist[list[int]]r   )r   s   "r   r   r   ´   s   € ÷ ñ ˜sð  ñ r   c                ó²  € \         P                  ! V 4      '       d.   V P                  4       P                  4       P	                  4       p M!\        V \        4      '       d   \        V 4      p \        V \        4      '       d	   V '       g   \        R 4      h\        ;QJ d    R V  4       F  '       d   K   RM	  RM! R V  4       4      '       d   \        V 4      .# . pV  Fõ  p\         P                  ! V4      '       d.   VP                  4       P                  4       P	                  4       pM!\        V\        4      '       d   \        V4      p\        V\        4      '       d	   V'       g   \        R4      h\        ;QJ d    R V 4       F  '       d   K   RM	  RM! R V 4       4      '       g   \        R4      hVP                  \        V4      4       K÷  	  V# )z1Token prompts must be a non-empty tensor or list.c              3  ó|   "  € T F2  p\        V\        4      ;'       d    \        V\        4      '       * x € K4  	  R # 5ir#   r,   ©r0   Útokens   & r   r2   Ú&_coerce_token_batch.<locals>.<genexpr>½   s,   é € Ð
UÉuÀeŒ:�eœSÓ!×AÐA¬*°U¼DÓ*AÔ&AÔAËuùr4   FTz8Each token prompt must be a non-empty list of token IDs.c              3  ó|   "  € T F2  p\        V\        4      ;'       d    \        V\        4      '       * x € K4  	  R # 5ir#   r,   r~   s   & r   r2   r€   È   s.   é € Ð[ÑWZÈe”:˜e¤SÓ)×IÐI´*¸UÄDÓ2IÔ.IÔIÓWZùr4   z1Token prompts may only contain integer token IDs.)r5   r6   r7   r8   r9   r   r;   r:   r%   r<   r=   Úappend)r)   ÚrowsÚrows   &  r   Ú_coerce_token_batchr…   ´   s>  € Ü‡‚�u×ÒØ—‘“×"Ñ"Ó$×+Ñ+Ó-‰Ü	�Eœ5×	!Ò	!Ü�U“ˆä�eœT×"Ò"¯%ÜÐLÓMÐMç
ƒsÑ
UÉuÓ
U‡s‡s‚sÑ
UÉuÓ
U×UÒUÜ�U“ˆ}Ðà€DÛˆÜ�?Š?˜3×ÒØ—*‘*“,×"Ñ"Ó$×+Ñ+Ó-‰CÜ˜œU×#Ò#Ü�s“)ˆCÜ˜#œt×$Ò$¯CÜÐWÓXÐXß‹sÑ[ÑWZÓ[�s�sŠsÑ[ÑWZÓ[×[Ò[ÜÐOÓPÐPØ�‰”D˜“IÖñ ð €Kr   c               ó$   € V ^8„  d   QhRRRRRR/# )r   Útoken_batchr{   Úattention_maskr   r   r   )r   s   "r   r   r   Î   s"   € ÷ ñ  ð Èð ÐP_ñ r   c                ó\  € Vf   V # \         P                  ! V4      '       d.   VP                  4       P                  4       P	                  4       pM!\        V\        4      '       d   \        V4      p\        V 4      ^8X  d\   \        V\        4      '       dF   V'       d>   \        ;QJ d    R V 4       F  '       d   K   RM	  RM! R V 4       4      '       d   V.p\        V\        4      '       d   \        V4      \        V 4      8w  d   \        R4      h. p\        W4       Fÿ  w  r4\         P                  ! V4      '       d.   VP                  4       P                  4       P	                  4       pM!\        V\        4      '       d   \        V4      p\        V\        4      '       d   \        V4      \        V4      8w  d   \        R4      h\        W44       UUu. uF  w  rV\        V4      '       g   K  VNK  	  pppV'       g   \        R4      hVP                  V4       EK  	  V# u uppi )Nc              3  óX   "  € T F   p\        V\        \        34      '       * x € K"  	  R # 5ir#   )r   r:   r;   r/   s   & r   r2   Ú(_apply_attention_mask.<locals>.<genexpr>×   s!   é € ÐN¹~°t”:˜d¤T¬5 MÓ2×2Ñ2»~ùs   ‚(*FTz>`attention_mask` must have the same batch size as `input_ids`.zHEach `attention_mask` row must have the same length as its token prompt.z3`attention_mask` removed every token from a prompt.)r5   r6   r7   r8   r9   r   r;   r:   Úlenr<   r%   Úzipr.   r‚   )r‡   rˆ   Úfiltered_batchÚ	token_idsÚmaskÚtoken_idÚkeepÚfiltereds   &&      r   Ú_apply_attention_maskr”   Î   s“  € ØÒØÐÜ‡‚�~×&Ò&Ø'×.Ñ.Ó0×4Ñ4Ó6×=Ñ=Ó?‰Ü	�N¤E×	*Ò	*Ü˜nÓ-ˆä
ˆ;Ó˜1Ô¤¨N¼D×!AÒ!AÇnß‹3ÑN¹~ÓN�3�3Š3ÑN¹~ÓN×NÒNØ,Ð-ˆNä�n¤d×+Ò+¬s°>Ó/BÄcÈ+ÓFVÔ/VÜÐYÓZÐZà€NÜ˜{Ö;‰ˆ	Ü�?Š?˜4× Ò Ø—;‘;“=×$Ñ$Ó&×-Ñ-Ó/‰DÜ˜œe×$Ò$Ü˜“:ˆDÜ˜$¤×%Ò%¬¨T«´c¸)³nÔ)DÜÐgÓhÐhÜ36°yÔ3GÔVÑ3G¡ Ì4ÐPTÏ:—H�HÑ3GˆÑVßÜÐRÓSÐSØ×Ñ˜h×'ñ <ð Ðùó	 Ws   ÇH(Ç4H(c               ó$   € V ^8„  d   QhRRRRRR/# )r   Úpromptsr   Ú	input_idsrˆ   r   )r   s   "r   r   r   ì   s&   € ÷ Dñ D Cð D°Cð DÈñ Dr   c                óÌ  € V e   Ve   \        R4      hV e   T MTpVf   \        R4      h\        V\        4      '       d   Ve   \        R4      hVR 3# \        V\        \
        34      '       d_   V'       dW   \        ;QJ d    R V 4       F  '       d   K   RM	  RM! R V 4       4      '       d   Ve   \        R4      h\	        V4      R 3# \        V4      pR \        WB4      3# )Nz*Pass only one of `prompts` or `input_ids`.z-Either prompts or input_ids must be provided.z`input_ids` cannot be a string.c              3  óB   "  € T F  p\        V\        4      x € K  	  R # 5ir#   )r   r   r/   s   & r   r2   Ú)_normalize_vllm_inputs.<locals>.<genexpr>÷   s   é € Ð9bÑ\aÐTX¼*ÀTÌ3×:OÐ:OÓ\aùs   ‚FTz#`input_ids` cannot contain strings.)	r%   r   r   r=   r:   r;   r<   r…   r”   )r–   r—   rˆ   r)   r‡   s   &&&  r   Ú_normalize_vllm_inputsr›   ì   sÐ   € ØÒ˜yÒ4ÜÐEÓFÐFØÒ*‰G°	€EØ‚}ÜÐHÓIÐIä�%œ×ÒØÒ ÜÐ=Ó>Ð>Ø�dˆ{ÐÜ�%œ$¤˜×'Ò'¯E·c³cÑ9bÑ\aÓ9b·c·c²cÑ9bÑ\aÓ9b×6bÒ6bØÒ ÜÐAÓBÐBÜ�E‹{˜DÐ Ð ä% eÓ,€KØÔ& {ÓCÐCÐCr   c               ó    € V ^8„  d   QhRRRR/# )r   r)   r   rl   zMapping[str, Any]r   )r   s   "r   r   r      s   € ÷ %^ñ %^ #ð %^Ð/@ñ %^r   c                ó„  € V fL   RpV Uu/ uF   q1P                  V4      f   K  W1V,          bK"  	  ppVP                  RR4       \        V4      # \        V \        4      '       d   V # \        V \
        4      '       d    \	        R/ \        V 4      B # \        R4      hu upi   \         d   p\        R4      ThR p?ii ; i)NrE   Fz'Invalid vLLM `sampling_params` mapping.zF`sampling_params` must be a vLLM SamplingParams instance or a mapping.)rF   rH   rI   rJ   rK   rL   rB   rC   rM   rN   rO   rP   rW   rQ   rR   rS   rT   rU   rV   rE   rD   r   )	r[   Ú
setdefaultri   r   r   r   r]   r=   r%   )r)   rl   Úhf_keysÚkeyr@   rg   s   &&    r   Ú_build_sampling_paramsr¡      sÀ   € Ø‚}ð
ˆñ. 29ÓX±¨#¿J¹JÀs»OÔ%�S �+Ò%±ˆ	ÐXð 	×Ñ˜[¨%Ô0Ü(¨Ó3Ð3Ü�%œ×(Ò(ØˆÜ�%œ×!Ò!ð	QÜ!Ñ0¤D¨£KÑ0Ð0ô Ð\Ó
]Ð]ùò Yøô ô 	QÜÐFÓGÈSÐPûð	Qús"   ‹B¤BÁ?B# Â#B?Â.B:Â:B?c                óÞ   € Vf   V P                   ! RRVRV/VB # \        e2   V Uu. uF  p\        VR7      NK  	  ppV P                   ! RRVRV/VB # V P                   ! RRVRV/VB # u upi )Nr–   Úsampling_params)Úprompt_token_idsr¤   r   )Úgenerater   )rk   Útext_promptsr‡   r£   Úgeneration_kwargsÚpromptÚtoken_promptss   &&&&&  r   Ú_run_vllm_generationrª   (  s„   € ØÒØ�~Š~Ñi lÐiÀOÐiÐWhÑiÐiÜÒÙMXÓYÉ[À6œ°v×>É[ˆÐYØ�~Š~Ñj mÐjÀ_ÐjÐXiÑjÐjØ�>Š>Ñm¨;ÐmÈÐmÐ[lÑmÐmùò Zs   §A*c           	     óÐ  € \        4        VP                  R R4      pVP                  RR4      pVP                  RR4      p\        W#V4      w  rV\        VP                  RR4      V4      pRpV U	u/ uF  q™V9   g   K  W‘P                  V	4      bK  	  p
p	\	        V VVVV
4      p. pV FV  p\        VP                  4      pVP                   F.  pVP                  V\        VP                  4      ,           4       K0  	  KX  	  V'       g'   \        P                  ! R	\        P                  R7      # VP                  R4      pVf-   V P                  4       pVP                  pVf   VP                   pVf   ^ p\#        R V 4       4      pV Uu. uF#  qÿV.V\%        V4      ,
          ,          ,           NK%  	  pp\        P&                  ! V\        P                  R7      # u up	i u upi )
r–   Nr—   rˆ   r£   )ÚdtypeÚpad_token_idc              3  ó8   "  € T F  p\        V4      x € K  	  R # 5ir#   )rŒ   )r0   Úoutputs   & r   r2   Ú vllm_generate.<locals>.<genexpr>\  s   é € Ð7©w V”S˜—[�[«wùs   ‚)Úuse_tqdmÚlora_requestÚpriorityÚtokenization_kwargsÚmm_processor_kwargs)r   r   )r'   Úpopr›   r¡   rª   r:   r¤   Úoutputsr‚   r�   r5   ÚemptyÚlongr[   Úget_tokenizerr­   rW   ÚmaxrŒ   Útensor)rk   rl   r–   r—   rˆ   r¦   r‡   r£   Úgenerate_keysr    r§   Úreq_resultsr·   Úresultr¤   r¯   r­   Ú	tokenizerrB   Úpaddeds   &,                  r   Úvllm_generaterÂ   1  s¸  € ä„Oà�j‰j˜ DÓ)€GØ—
‘
˜;¨Ó-€IØ—Z‘ZÐ 0°$Ó7€NÜ 6°wÈ>Ó ZÑ€Lä,¨V¯Z©ZÐ8IÈ4Ó-PÐRXÓY€Oð€Mñ :GÓX¹°#ÐQWÉ-Ô-˜Ÿj™j¨›oÒ-¹ÐÐXÜ&ØØØØØó€Kð €GÛˆÜ × 7Ñ 7Ó8ÐØ—n”nˆFØ�N‰NÐ+¬d°6×3CÑ3CÓ.DÕDÖEó %ñ ÷ Ü�{Š{˜6¬¯©Ô4Ð4à—:‘:˜nÓ-€LØÒØ×'Ñ'Ó)ˆ	Ø ×-Ñ-ˆØÒØ$×1Ñ1ˆLØÒØˆäÑ7©wÓ7Ó7€JÙQXÓYÑQXÀv˜�~¨´c¸&³kÕ)AÕB×BÐBÑQX€FÐYÜ�<Š<˜¤e§j¡jÔ1Ð1ùò9 Yùò6 Zs   Á1GÁ>GÆ)G#)*Ú__conditional_annotations__Ú
__future__r   Úimportlib.metadatar   r   Útypingr   r   r   r	   r5   r
   Ú__annotations__r   r$   r   r   r   r   Ú_LLMÚ_SamplingParamsÚ_TokensPromptÚImportErrorÚvllm.inputsÚ	Exceptionrg   r   r&   r'   r>   ri   rm   ru   ry   r…   r”   r›   r¡   rª   Úinference_moderÂ   )rÃ   s   @r   Ú<module>rÏ      sA  øð÷ #Ð "ç <ß /Ó /ã ð #€ˆmÓ "Ø)-Ð Ð&Ó -Ø€à
€Ø€Ø€ð	Ù˜6“?€LðÝ Ý6ð!Ý6ð €CØ$€NØ €LØ€Nõ
	Nñ .Ó/Ð õCõ
	Tõ8òvòò$	2õõ4õ<Dõ(%^òPnð ×ÒÓñ,2ó ò,2øðu ô 	Ùð	ûð ô !ð	!ßAøØô 	!Ø ‹Mð	!úð!ûð ô ØÕûðúsk   ºB: ÁC+ ÁC ÁC+ Â:CÃCÃC(ÃCÃC+ Ã	C$Ã C(Ã!C+ Ã#C$Ã$C(Ã(C+ Ã+C>Ã2C9Ã9C>