+
    œ°³j1  ã                   ó|   € R t ^ RIHt ^RIHt ^RIHtHt ^RIH	t	 R.t
]	! RR7      ] ! R R]4      4       4       tR	# )
z#RADIO vision encoder configuration.)Ústrict)ÚPreTrainedConfig)ÚOPENAI_CLIP_MEANÚOPENAI_CLIP_STD)Úauto_docstringÚRadioConfigznvidia/C-RADIOv4-H)Ú
checkpointc                   óÒ   a a€ ] tR t^t oRtRtRt^ t^tRt	Rt
RtRtRtRtRtR	tR
t^t^t^àtRt^t^tRt]! ]4      t]! ]4      tRtV 3R lt] V3R lR l4       t!V3R lt"Rt#Vt$V ;t%# )r   ax  
mlp_ratio (`float`, *optional*, defaults to 4.0):
    Ratio of the hidden size of the MLP relative to `hidden_size`.
use_swiglu_ffn (`bool`, *optional*, defaults to `False`):
    Whether to use a SwiGLU feed-forward network in the encoder layers instead of the standard MLP.
layerscale_value (`float`, *optional*, defaults to 1.0):
    Initial value for the LayerScale parameters. C-RADIO has no LayerScale; the default of `1.0` makes the
    (inherited) LayerScale an identity operation.
max_img_size (`int`, *optional*, defaults to 2048):
    Maximum supported image size (in pixels) used to size the position embedding table of the CPE patch generator.
num_cls_tokens (`int`, *optional*, defaults to 3):
    Number of learned class (summary) tokens prepended to the patch sequence.
num_registers (`int`, *optional*, defaults to 7):
    Number of learned register tokens prepended to the patch sequence.
summary_idxs (`list[int]`, *optional*, defaults to `[0, 1]`):
    Indices of the class tokens to gather and flatten into the `summary` output embedding.
norm_mean (`tuple[float, float, float]`, *optional*, defaults to `OPENAI_CLIP_MEAN`):
    Per-channel mean used by the input conditioner to normalize pixel values.
norm_std (`tuple[float, float, float]`, *optional*, defaults to `OPENAI_CLIP_STD`):
    Per-channel standard deviation used by the input conditioner to normalize pixel values.
Úradioi   g      @Úgelug�íµ ÷Æ°>g        FTg      ð?i   Ng{®Gáz”?c                ó¾   <€ V P                   f
   ^ ^.V n         \        V P                  4      V n        \        V P                  4      V n        \        SV `  ! R/ VB  R # )N© )Úsummary_idxsÚlistÚ	norm_meanÚnorm_stdÚsuperÚ__post_init__)ÚselfÚkwargsÚ	__class__s   &,€Ún/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/radio/configuration_radio.pyr   ÚRadioConfig.__post_init__P   sK   ø€ Ø×ÑÒ$Ø!" A ˆDÔÜ˜dŸn™nÓ-ˆŒÜ˜TŸ]™]Ó+ˆŒÜ‰ÒÑ' Ô'ó    c                ó    <€ V ^8„  d   QhRS[ /# )é   Úreturn)Úint)ÚformatÚ__classdict__s   "€r   Ú__annotate__ÚRadioConfig.__annotate__X   s   ø€ ÷ 8ñ 8¡Cñ 8r   c                ó<   € V P                   V P                  ,           # )zJNumber of skipped prefix tokens (cls + registers) before spatial features.)Únum_cls_tokensÚnum_registers)r   s   &r   Únum_summary_tokensÚRadioConfig.num_summary_tokensW   s   € ð ×"Ñ" T×%7Ñ%7Õ7Ð7r   c                ó´  <€ V ^8„  d   Qh/ S[ ;R&   S[ ;R&   S[ ;R&   S[;R&   S[;R&   S[;R&   S[;R&   S[;R&   S[;R	&   S[;R
&   S[;R&   S[;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[S[ ,          R,          ;R&   S[S[,          S[S[S[S[3,          ,          ;R&   S[S[,          S[S[S[S[3,          ,          ;R&   S[;R&   # )r   Úhidden_sizeÚnum_hidden_layersÚnum_attention_headsÚ	mlp_ratioÚ
hidden_actÚlayer_norm_epsÚattention_probs_dropout_probÚhidden_dropout_probÚdrop_path_rateÚuse_swiglu_ffnÚqkv_biasÚlayerscale_valueÚnum_channelsÚ
patch_sizeÚ
image_sizeÚmax_img_sizer#   r$   Nr   r   r   Úinitializer_range)r   ÚfloatÚstrÚboolr   Útuple)r   r   s   "€r   r    r!      sx  ø‡ ‚ ñ6 Ññ7 ñ8 Ññ9 ñ: Ñ!ñ; ñ< Ññ= ñ> Ññ? ñ@ Ñ ñA ñB #(Ñ-ñC ñD Ñ$ñE ñF ÑñG ñH Ñ ñI ñJ ÑñK ñN Ñ!ñO ñR ÑñS ñT ÑñU ñV ÑñW ñZ Ññ[ ñ\ Ññ] ñ^ Ññ_ ñ` ‘s•)˜dÕ"Ñ)ña ñd ‘E�{™U¡5©%±Ð#6Õ7Õ7ÑQñe ñf ‘5�k™E¡%©±Ð"5Õ6Õ6ÑOñg ñh Ñ#òi r   )r   r   r   )&Ú__name__Ú
__module__Ú__qualname__Ú__firstlineno__Ú__doc__Ú
model_typer(   r)   r*   r+   r,   r-   r.   r/   r0   r1   r2   r3   r4   r5   r6   r7   r#   r$   r   r<   r   r   r   r   r8   r   Úpropertyr%   Ú__annotate_func__Ú__static_attributes__Ú__classdictcell__Ú__classcell__)r   r   s   @@r   r   r      s±   ù‡ € ñð, €Jà€KØÐØ!ÐØ€IØ€JØ €NØ*-Ð Ø!$ÐØ€NØ €NØ€Hà!Ðà€LØ€JØ€Jà€LØ€NØ€MØ%)€Lá:?Ð@PÓ:Q€IÙ9>¸Ó9O€HØ#Ðõ(ð ÷8ó ð8÷} … r   N)rA   Úhuggingface_hub.dataclassesr   Úconfiguration_utilsr   Úimage_utilsr   r   Úutilsr   Ú__all__r   r   r   r   Ú<module>rM      sI   ðñ *å .å 3ß <Ý #ð ˆ/€ñ Ð/Ô0Øô>8Ð"ó >8ó ó 1ò>8r   