+
    œ°³j®6  ã                   ó  € ^ RI Ht ^RIHt ^RIHt ]! RR7      ] ! R R]4      4       4       t]! RR7      ] ! R R	]4      4       4       t]! RR7      ] ! R
 R]4      4       4       t]! RRR7      ] ! R R]4      4       4       t	. ROt
R# )é    )Ústrict)ÚPreTrainedConfig)Úauto_docstringzthinkingmachines/Inkling)Ú
checkpointc                   ó�  a a€ ] tR t^t oRtRtRtRRRRRR	R
RRRRRRR	RRRRRRRR	/tRR.R.3RRR.R.3RR.R.3/tRRRRRRR
R/t	RRRR R!R"R#R$R%R&/t
R'tR(tR)t^Bt^@t^t^€t^@t^t^€tR*t^tR+tR(tR,tR(tR(tR-tR.t^tR(tR/t R0t!R1t"R2t#^t$^t%R3t&R4t'R5t(R.t)R6t*R7t+R(t,^t-^t.R(t/R8t0R3t1R(t2V 3R9 lt3]4R: 4       t5]4R; 4       t6V3R< lt7R=t8Vt9V ;t:# )>ÚInklingTextConfiga$  
unpadded_vocab_size (`int`, *optional*, defaults to `None`):
    Number of rows the checkpoint's unembedding matrix actually holds when the head is not padded to
    `vocab_size`. Logits beyond it are dropped. If `None`, the head is not padded.
swa_num_attention_heads (`int`, *optional*, defaults to 64):
    Number of attention heads in the sliding-window layers.
swa_num_key_value_heads (`int`, *optional*, defaults to 16):
    Number of key/value heads in the sliding-window layers.
swa_head_dim (`int`, *optional*, defaults to 128):
    Dimension of query and key heads in the sliding-window layers.
sliding_window_size (`int`, *optional*, defaults to 512):
    Size of the sliding attention window used by layers whose `layer_types` entry is `"hybrid_sliding"`.
d_rel (`int`, *optional*, defaults to 16):
    Per-head dimension of the relative states that are mixed into the relative position bias.
rel_extent (`int`, *optional*, defaults to 1024):
    Backward distance, in tokens, over which the relative position bias is applied. The bias is zero beyond it.
log_scaling_n_floor (`int`, *optional*, defaults to `None`):
    Position from which logits start being scaled up logarithmically in the full-attention layers. If `None`,
    the scaling is disabled.
log_scaling_alpha (`float`, *optional*, defaults to 0.1):
    Strength of the logarithmic logit scaling controlled by `log_scaling_n_floor`.
local_layer_ids (`list[int]`, *optional*, defaults to `None`):
    Indices of the layers using sliding window attention. Used to derive `layer_types` when it is not provided.
    If `None`, every layer whose index is not a multiple of 6 uses sliding window attention.
mlp_layer_types (`list[str]`, *optional*, defaults to `None`):
    MLP type pattern for each layer (`"dense"` or `"sparse"`). If `None`, every layer is sparse.
shared_expert_sink (`bool`, *optional*, defaults to `True`):
    Whether the router scores the shared experts alongside the routed ones, so that they act as a sink in the
    softmax over expert weights.
logits_mup_width_multiplier (`float`, *optional*, defaults to 24.0):
    muP width multiplier the final hidden states are divided by before the language modeling head.
rms_norm_eps_moe_gate (`float`, *optional*, defaults to 1e-6):
    Epsilon of the RMS normalization applied inside the mixture-of-experts router.
num_mtp_layers (`int`, *optional*, defaults to `None`):
    Number of multi-token-prediction layers. If `None`, multi-token prediction is disabled.
chain_hidden_post_norm (`bool`, *optional*, defaults to `False`):
    Whether the hidden states chained between multi-token-prediction layers are normalized after each layer.
mtp_hidden_states_first (`bool`, *optional*, defaults to `True`):
    Whether the hidden states come before the token embeddings when the two are concatenated as the input of a
    multi-token-prediction layer.
mtp_local_layer_ids (`list[int]`, *optional*, defaults to `None`):
    Indices of the multi-token-prediction layers using sliding window attention. If `None`, every
    multi-token-prediction layer uses full attention.
Úinkling_textÚtext_configÚembed_tokensÚembedding_rowwisez!layers.*.mlp.experts.gate_up_projÚcolwisezlayers.*.mlp.experts.down_projÚrowwisezlayers.*.mlp.expertsÚmoe_tp_expertsz%layers.*.mlp.shared_experts.gate_projz#layers.*.mlp.shared_experts.up_projz%layers.*.mlp.shared_experts.down_projzlayers.*.mlp.shared_expertsÚ
all_reducezlayers.*.mlp.gate_projzlayers.*.mlp.up_projzlayers.*.mlp.down_projÚ	input_idsÚinputs_embedsÚlayersÚhidden_statesÚattention_maskÚnormzlayers.*.mlp.gateÚ	ep_routerÚgrouped_gemmÚembedding_multiplierÚlogits_mup_width_multiplierÚsliding_windowÚsliding_window_sizeÚnum_local_expertsÚn_routed_expertsÚsconv_kernel_sizeÚconv_kernel_sizeÚmodel_max_lengthÚmax_position_embeddingsi@ Né   i   é   gš™™™™™¹?i   ç�íµ ÷Æ°>i `  Úsilui   é   Tg       @g      8@g        ç{®Gáz”?Fc                óˆ  <€ V P                   f’   V P                  e   \        V P                  4      pM;\        V P                  4       Uu0 uF  q3^,           ^,          '       g   K  VkK  	  pp\        V P                  4       Uu. uF  q3V9   d   RMRNK  	  upV n         V P
                  fE   VP                  R^ 4      p\        V P                  4       Uu. uF  q3V8  d   RMRNK  	  upV n        VP                  R4      e   VP                  R4      V n        ^V n	        \        SV `,  ! R/ VB  R # u upi u upi u upi )NÚhybrid_slidingÚhybridÚdense_mlp_idxÚdenseÚsparseÚdense_intermediate_size© )Úlayer_typesÚlocal_layer_idsÚsetÚrangeÚnum_hidden_layersÚmlp_layer_typesÚpopÚgetÚintermediate_sizeÚnumber_of_conv_statesÚsuperÚ__post_init__)ÚselfÚkwargsr2   Úir,   Ú	__class__s   &,   €Úr/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/inkling/configuration_inkling.pyr<   ÚInklingTextConfig.__post_init__Ÿ   s*  ø€ Ø×ÑÒ#Ø×#Ñ#Ò/Ü"% d×&:Ñ&:Ó";‘ä.3°D×4JÑ4JÔ.KÓ"[Ñ.K¨ÐTUÕPUÐYZÏ{É{§1 1Ñ.K�Ð"[äPUÐVZ×VlÑVlÔPmó ÙPmÈ1¨Ô$8Ñ ¸hÒFÑPmñ ˆDÔð ×ÑÒ'Ø"ŸJ™J ¸Ó:ˆMÜX]Ð^b×^tÑ^tÔXuÓ#vÑXuÐST°=Ô/@¡GÀhÒ$NÑXuÑ#vˆDÔ à�:‰:Ð/Ó0Ò<Ø%+§Z¡ZÐ0IÓ%JˆDÔ"ð &'ˆÔ"ä‰ÒÑ' Ô'ùò #\ùò ùò
 $ws   ÁD5Á'D5ÂD:ÃD?c                óÞ   € V P                   eZ   V P                  f   R.V P                   ,          # \        V P                   4       Uu. uF  qV P                  9   d   RMRNK  	  up# R # u upi )Nr+   r*   )Únum_mtp_layersÚmtp_local_layer_idsr4   )r=   r?   s   & rA   Úmtp_layer_typesÚ!InklingTextConfig.mtp_layer_types´   sv   € à×ÑÒ*Ø×'Ñ'Ò/Ø �z D×$7Ñ$7Õ7Ð7ô ^cÐcg×cvÑcvÔ]wóÙ]wÐXY¨T×-EÑ-EÔ(EÑ$È8ÒSÑ]wñð ñ ùòs   ÁA*c                óJ   € V P                   e   R.V P                   ,          # R # )Nr-   )rD   )r=   s   &rA   Úmtp_mlp_layer_typesÚ%InklingTextConfig.mtp_mlp_layer_typesÀ   s$   € à×ÑÒ*Ø�9˜t×2Ñ2Õ2Ð2Ùó    c                óÆ  <€ V ^8„  d   Qh/ S[ ;R&   S[ R,          ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R	&   S[ ;R
&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ R,          ;R&   S[;R&   S[S[ ,          R,          ;R&   S[S[,          R,          ;R&   S[ ;R&   S[;R&   S[ ;R&   S[S[,          R,          ;R&   S[ ;R&   S[;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[;R&   S[;R&   S[;R&   S[;R &   S[;R!&   S[;R"&   S[ R,          ;R#&   S[ R,          ;R$&   S[ R,          ;R%&   S[ R,          ;R&&   S[;R'&   S[;R(&   S[S[ ,          R,          ;R)&   # )*é   Ú
vocab_sizeNÚunpadded_vocab_sizeÚhidden_sizer5   Únum_attention_headsÚnum_key_value_headsÚhead_dimÚswa_num_attention_headsÚswa_num_key_value_headsÚswa_head_dimr   Úd_relÚ
rel_extentÚlog_scaling_n_floorÚlog_scaling_alphar2   r1   r"   Úrms_norm_epsr    r6   r9   Ú
hidden_actÚmoe_intermediate_sizer   Únum_experts_per_tokÚn_shared_expertsÚshared_expert_sinkÚroute_scaler   Úrms_norm_eps_moe_gateÚattention_dropoutÚinitializer_rangeÚpad_token_idÚbos_token_idÚeos_token_idrD   Úchain_hidden_post_normÚmtp_hidden_states_firstrE   )ÚintÚfloatÚlistÚstrÚbool)ÚformatÚ__classdict__s   "€rA   Ú__annotate__ÚInklingTextConfig.__annotate__   s˜  ø‡ ‚ ñl Ññm ñp ˜t�Ñ*ñq ñr Ññs ñt Ññu ñv Ñ!ñw ñx Ñ ñy ñz Ññ{ ñ| !Ñ%ñ} ñ~ !Ñ%ñ ñ@ ÑñA ñB Ñ"ñC ñD �OñE ñF ÑñG ñH ˜t�Ñ*ñI ñJ Ñ"ñK ñL ™#•Y Õ%Ñ,ñM ñN ‘c•˜TÕ!Ñ(ñO ñP !Ñ)ñQ ñR ÑñS ñT ÑñU ñV ™#•Y Õ%Ñ,ñW ñX Ñ"ñY ñZ Ññ[ ñ^ Ñ%ñ_ ñ` Ñña ñb Ñ ñc ñd Ññe ñf Ñ#ñg ñh Ññi ñl "'Ñ-ñm ñn !Ñ'ño ñp Ñ"ñq ñr Ñ#ñs ñt ˜•*Ñ#ñu ñv ˜•*Ñ ñw ñx ˜•*Ñ ñy ñ| ˜$•JÑ%ñ} ñ~ !Ñ(ñ ñ@ "Ñ(ñA ñB ™c� TÕ)Ñ0òC rK   )r9   r1   r6   r:   );Ú__name__Ú
__module__Ú__qualname__Ú__firstlineno__Ú__doc__Ú
model_typeÚbase_config_keyÚbase_model_tp_planÚbase_model_pp_planÚbase_model_ep_planÚattribute_maprN   rO   rP   r5   rQ   rR   rS   rT   rU   rV   r   rW   rX   rY   rZ   r2   r1   r"   r[   r    r6   r9   r\   r]   r   r^   r_   r`   ra   r   rb   rc   rd   re   rf   rg   rD   rh   ri   rE   r<   ÚpropertyrF   rI   Ú__annotate_func__Ú__static_attributes__Ú__classdictcell__Ú__classcell__©r@   rp   s   @@rA   r   r      só  ù‡ € ñ+ðZ  €JØ#€OàÐ+Ø+¨YØ(¨)ØÐ 0Ø/°Ø-¨yØ/°Ø% |Ø  )Ø 	Ø  )ðÐð 	˜+˜¨Ð(9Ð:Ø�OÐ%5Ð6¸Ð8IÐJØ�/Ð" _Ð$5Ð6ðÐð 	˜[Ø+¨^Ø(¨.ØÐ 0ð	Ðð 	Ð =ØÐ/ØÐ/ð 	Ð/àÐ5ð	€Mð €Jà&*ÐØ€KØÐØ!ÐØ ÐØ€HØ#%ÐØ#%ÐØ€LØ"ÐØ€EØ€JØ&*ÐØ"ÐØ(,€OØ$(€KØ#)ÐØ€LØÐØ(,€OØ"ÐØ€Jà!%ÐØÐØ ÐØÐØ#ÐØ€Kà)-ÐØ#'ÐØ"ÐØ#ÐØ#€LØ €LØ €Là!%€NØ#(ÐØ$(ÐØ,0Ðõ(ð* ñó ðð ñó ð÷K … rK   r   c                   ó\   a € ] tR t^Çt o RtRtRtRRRRRR	/t^PtR
t	Rt
RtRtV 3R ltRtV tR# )ÚInklingAudioConfigaz  
n_mel_bins (`int`, *optional*, defaults to 80):
    Number of mel-frequency bins per audio frame.
mel_vocab_size (`int`, *optional*, defaults to 256):
    Number of discrete bins each mel value is quantized into before being embedded.
text_hidden_size (`int`, *optional*, defaults to 6144):
    Dimensionality the audio embeddings are projected to, matching the text backbone.
Úinkling_audioÚaudio_configÚnum_codebooksÚ
n_mel_binsÚcodebook_sizeÚmel_vocab_sizerP   Útext_hidden_sizer'   r#   r%   r(   c                óV   <€ V ^8„  d   Qh/ S[ ;R&   S[ ;R&   S[ ;R&   S[;R&   S[;R&   # )rM   r‰   r‹   rŒ   r[   rd   ©rj   rk   )ro   rp   s   "€rA   rq   ÚInklingAudioConfig.__annotate__Ç   sF   ø‡ ‚ ñ( Ññ) ñ* Ññ+ ñ, Ñ ñ- ñ. Ññ/ ñ0 Ñ#ò1 rK   r0   N)rs   rt   ru   rv   rw   rx   ry   r}   r‰   r‹   rŒ   r[   rd   r   r€   r�   ©rp   s   @rA   r…   r…   Ç   sR   ø‡ € ñð !€JØ$€Oà˜ØÐ)ØÐ)ð€Mð €JØ€NØ ÐØ€LØ#Ð÷1 ƒ rK   r…   c                   ód   a € ] tR t^ât o RtRtRtRR/tRt^(t	^t
^tRt^t^tRtR	tV 3R
 ltRtV tR# )ÚInklingVisionConfigz¨
text_hidden_size (`int`, *optional*, defaults to 6144):
    Dimensionality the vision features are projected to by the last encoder layer, matching the text backbone.
Úinkling_visionÚvision_configr5   Ún_layersr#   r$   r%   r(   c                ó†   <€ V ^8„  d   Qh/ S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[;R&   S[;R	&   # )
rM   rŒ   Ú
patch_sizeÚtemporal_patch_sizeÚnum_channelsrP   r5   rQ   r[   rd   rŽ   )ro   rp   s   "€rA   rq   Ú InklingVisionConfig.__annotate__â   sz   ø‡ ‚ ñ Ñ ñ ñ Ññ ñ Ñ ñ ñ Ññ ñ  Ññ! ñ" Ññ# ñ$ Ñ!ñ% ñ& Ññ' ñ( Ñ#ò) rK   r0   N)rs   rt   ru   rv   rw   rx   ry   r}   rŒ   r—   r˜   r™   rP   r5   rQ   r[   rd   r   r€   r�   r�   s   @rA   r’   r’   â   sY   ø‡ € ñð
 "€JØ%€OØ(¨*Ð5€Mà ÐØ€JØ ÐØ€LØ€KØÐØ!ÐØ€LØ#Ð÷) ƒ rK   r’   zETop-level multimodal config (`InklingMMConfig` in the SGLang source).)r   Úcustom_introc                   ór   a a€ ] tR t^ùt oRtRtR]R]R]/t	Rt
RtRtRtRtR	tR
tV 3R ltV3R ltRtVtV ;t# )ÚInklingConfiga  
image_bos_token_id (`int`, *optional*, defaults to 200005):
    The beginning-of-image token index used to mark the start of image spans.
audio_bos_token_id (`int`, *optional*, defaults to 200020):
    The beginning-of-audio token index used to mark the start of audio spans.
Úinkling_mm_modelr
   r‡   r”   Niv iu iE iT c                óî  <€ VP                  R 4      ;'       g    / p\        V P                  \        4      '       dƒ   V P                  P	                  RVP                  R4      4       V P                  P	                  RVP                  RR4      4       V P                  P	                  RVP                  R4      4       \        V P
                  \        4      '       d*   V P                  R,          ! R/ V P
                  B V n        M+V P
                  f   V P                  R,          ! 4       V n        \        V P                  \        4      '       d*   V P                  R	,          ! R/ V P                  B V n        M+V P                  f   V P                  R	,          ! 4       V n        \        V P                  \        4      '       d*   V P                  R
,          ! R/ V P                  B V n        M+V P                  f   V P                  R
,          ! 4       V n        V P                  P                  V P                  n	        V P                  P                  V P
                  n	        \        SV `,  ! R/ VB  R# )Ú
mtp_configrD   Únum_nextn_predict_layersrh   FrE   r2   r‡   Nr”   r
   r0   )r8   Ú
isinstancer
   ÚdictÚ
setdefaultr‡   Úsub_configsr”   rP   rŒ   r;   r<   )r=   r>   r    r@   s   &, €rA   r<   ÚInklingConfig.__post_init__  sØ  ø€ à—Z‘Z Ó-×3Ð3°ˆ
Ü�d×&Ñ&¬×-Ò-Ø×Ñ×'Ñ'Ð(8¸*¿.¹.ÐIcÓ:dÔeØ×Ñ×'Ñ'Ð(@À*Ç.Á.ÐQiÐkpÓBqÔrØ×Ñ×'Ñ'Ð(=¸z¿~¹~ÐN_Ó?`Ôaä�d×'Ñ'¬×.Ò.Ø $× 0Ñ 0°Ö @Ñ UÀ4×CTÑCTÑ UˆDÕØ×ÑÒ&Ø $× 0Ñ 0°Ö @Ó BˆDÔä�d×(Ñ(¬$×/Ò/Ø!%×!1Ñ!1°/Ö!BÑ!XÀT×EWÑEWÑ!XˆDÕØ×ÑÒ'Ø!%×!1Ñ!1°/Ö!BÓ!DˆDÔä�d×&Ñ&¬×-Ò-Ø#×/Ñ/°Ö>ÑRÀ×AQÑAQÑRˆDÕØ×ÑÒ%Ø#×/Ñ/°Ö>Ó@ˆDÔà.2×.>Ñ.>×.JÑ.Jˆ×ÑÔ+Ø-1×-=Ñ-=×-IÑ-Iˆ×ÑÔ*Ü‰ÒÑ' Ô'rK   c                óÈ   <€ V ^8„  d   Qh/ S[ S[,          R,          ;R&   S[S[,          R,          ;R&   S[S[,          R,          ;R&   S[;R&   S[;R&   S[;R&   S[;R&   # )	rM   Nr
   r‡   r”   Úimage_token_idÚaudio_token_idÚimage_bos_token_idÚaudio_bos_token_id)r   r£   r…   r’   rj   )ro   rp   s   "€rA   rq   ÚInklingConfig.__annotate__ù   s~   ø‡ ‚ ñ( #¡TÕ)¨DÕ0Ñ7ñ) ñ* %¡tÕ+¨dÕ2Ñ9ñ+ ñ, '©Õ-°Õ4Ñ;ñ- ñ. Ñ ñ/ ñ0 Ñ ñ1 ñ2 Ñ$ñ3 ñ4 Ñ$ò5 rK   )r‡   r
   r”   )rs   rt   ru   rv   rw   rx   r   r…   r’   r¥   r
   r‡   r”   r¨   r©   rª   r«   r<   r   r€   r�   r‚   rƒ   s   @@rA   r�   r�   ù   s]   ù‡ € ñð $€JàÐ(ØÐ*ØÐ,ð€Kð 48€KØ59€LØ7;€MØ €NØ €NØ$ÐØ$Ðõ(÷9 … rK   r�   N)r�   r   r…   r’   )Úhuggingface_hub.dataclassesr   Úconfiguration_utilsr   Úutilsr   r   r…   r’   r�   Ú__all__r0   rK   rA   Ú<module>r±      sÌ   ðõ, /å 3Ý #ñ Ð5Ô6ØôfÐ(ó fó ó 7ðfñR Ð5Ô6Øô$Ð)ó $ó ó 7ð$ñ2 Ð5Ô6Øô$Ð*ó $ó ó 7ð$ñ* Ø)ØXôð ô0(Ð$ó 0(ó ó	ð
0(òf ^‚rK   