+
    œ°³jÓ  ã                   ól   € R t ^ RIHt ^RIHt ^RIHt ]! RR7      ] ! R R]4      4       4       tR.tR# )	zLongT5 model configuration)Ústrict)ÚPreTrainedConfig)Úauto_docstringzgoogle/long-t5-local-base)Ú
checkpointc                   óÂ   a a€ ] tR t^t oRtRtR.tRRRRRR	R
R/tRtRt	^@t
Rt^tRt^t^t^t^ t^€tRtRtRtRtRtRtRt^ t^tRtRtRtV 3R ltR t V3R lt!Rt"Vt#V ;t$# )ÚLongT5ConfigaÙ  
d_ff (`int`, *optional*, defaults to 2048):
    Size of the intermediate feed forward layer in each `LongT5Block`.
local_radius (`int`, *optional*, defaults to 127):
    Number of tokens to the left/right for each token to locally self-attend in a local attention mechanism.
global_block_size (`int`, *optional*, defaults to 16):
    Length of blocks an input sequence is divided into for a global token representation. Used only for
    `encoder_attention_type = "transient-global"`.
relative_attention_num_buckets (`int`, *optional*, defaults to 32):
    The number of buckets to use for each attention layer.
relative_attention_max_distance (`int`, *optional*, defaults to 128):
    The maximum distance of the longer sequences for the bucket separation.
feed_forward_proj (`string`, *optional*, defaults to `"relu"`):
    Type of feed forward layer to be used. Should be one of `"relu"` or `"gated-gelu"`. LongT5v1.1 uses the
    `"gated-gelu"` feed forward projection. Original LongT5 implementation uses `"gated-gelu"`.
encoder_attention_type (`string`, *optional*, defaults to `"local"`):
    Type of encoder attention to be used. Should be one of `"local"` or `"transient-global"`, which are
    supported by LongT5 implementation.
Úlongt5Úpast_key_valuesÚhidden_sizeÚd_modelÚnum_attention_headsÚ	num_headsÚnum_hidden_layersÚ
num_layersÚhead_dimÚd_kvi€}  i   i   Ngš™™™™™¹?g�íµ ÷Æ°>g      ð?ÚreluTÚlocalFc                óR  <€ V P                   e   V P                   MV P                  V n         V P                  P                  R4      pVR,          V n        V^ ,          R8H  V n        V P                  R8X  d   RV n        V P                  V n        RV n        \        SV `$  ! R/ VB  R # )NÚ-Úgatedz
gated-geluÚgelu_newTéÿÿÿÿ© )
Únum_decoder_layersr   Úfeed_forward_projÚsplitÚdense_act_fnÚis_gated_actÚtie_word_embeddingsÚscale_decoder_outputsÚsuperÚ__post_init__)ÚselfÚkwargsÚact_infoÚ	__class__s   &, €Úp/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/longt5/configuration_longt5.pyr"   ÚLongT5Config.__post_init__N   s–   ø€ Ø=A×=TÑ=TÒ=` $×"9Ò"9Ðfj×fuÑfuˆÔØ×)Ñ)×/Ñ/°Ó4ˆØ$ R�LˆÔØ$ Q�K¨7Ñ2ˆÔà×!Ñ! \Ô1Ø *ˆDÔð &*×%=Ñ%=ˆÔ"Ø#'ˆÔ ä‰ÒÑ' Ô'ó    c                óÊ   € V P                   P                  R4      p\        V4      ^8”  d   V^ ,          R8w  g   \        V4      ^8”  d   \        RV P                    R24      hR# )zOPart of `@strict`-powered validation. Validates the architecture of the config.r   r   z`feed_forward_proj`: z© is not a valid activation function of the dense layer. Please make sure `feed_forward_proj` is of the format `gated-{ACT_FN}` or `{ACT_FN}`, e.g. 'gated-gelu' or 'relu'N)r   r   ÚlenÚ
ValueError)r#   r%   s   & r'   Úvalidate_architectureÚ"LongT5Config.validate_architecture^   sf   € à×)Ñ)×/Ñ/°Ó4ˆÜˆx‹=˜1Ô ¨!¥°Ô!7¼3¸x»=È1Ô;LÜØ'¨×(>Ñ(>Ð'?ð @)ð )óð ñ <Mr)   c                ó–  <€ V ^8„  d   Qh/ S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ ;R&   S[ R,          ;R&   S[ ;R&   S[ ;R	&   S[ ;R
&   S[ ;R&   S[ ;R&   S[S[ ,          ;R&   S[;R&   S[;R&   S[;R&   S[;R&   S[;R&   S[;R&   S[ R,          ;R&   S[ S[S[ ,          ,          R,          ;R&   S[ R,          ;R&   S[;R&   S[;R&   # )é   Ú
vocab_sizer   r   Úd_ffr   Nr   r   Úlocal_radiusÚglobal_block_sizeÚrelative_attention_num_bucketsÚrelative_attention_max_distanceÚdropout_rateÚlayer_norm_epsilonÚinitializer_factorr   Úis_encoder_decoderÚencoder_attention_typeÚ	use_cacheÚpad_token_idÚeos_token_idÚbos_token_idÚ
is_decoderr   )ÚintÚfloatÚstrÚboolÚlist)ÚformatÚ__classdict__s   "€r'   Ú__annotate__ÚLongT5Config.__annotate__   sz  ø‡ ‚ ñ@ ÑñA ñB ÑñC ñD �NñE ñF ÑñG ñH ÑñI ñJ ˜d�
Ñ)ñK ñL ÑñM ñN ÑñO ñP ÑñQ ñR %(Ñ,ñS ñT &)Ñ.ñU ñV ™#•+Ñ#ñW ñX Ñ$ñY ñZ Ñ#ñ[ ñ\ Ñ#ñ] ñ^ Ñ#ñ_ ñ`  Ñ)ña ñb Ññc ñd ˜•*Ñ ñe ñf ™™S�	•/ DÕ(Ñ,ñg ñh ˜•*Ñ#ñi ñj Ññk ñl Ñ$òm r)   )r   r   r   r    r   )%Ú__name__Ú
__module__Ú__qualname__Ú__firstlineno__Ú__doc__Ú
model_typeÚkeys_to_ignore_at_inferenceÚattribute_mapr1   r   r   r2   r   r   r   r3   r4   r5   r6   r7   r8   r9   r   r:   r;   r<   r=   r>   r?   r@   r   r"   r-   Ú__annotate_func__Ú__static_attributes__Ú__classdictcell__Ú__classcell__)r&   rG   s   @@r'   r   r      sÆ   ù‡ € ñð( €JØ#4Ð"5Ðà�yØ˜{Ø˜\Ø�Fð	€Mð €JØ€GØ€DØ€DØ€JØ%)ÐØ€IØ€LØÐØ*,Ð"Ø+.Ð#Ø #€LØ $ÐØ #ÐØ#ÐØ#ÐØ")ÐØ€IØ €LØ+,€LØ#€LØ€JØ $Ðõ(ò ÷Q … r)   r   N)	rN   Úhuggingface_hub.dataclassesr   Úconfiguration_utilsr   Úutilsr   r   Ú__all__r   r)   r'   Ú<module>rZ      sK   ðñ !å .å 3Ý #ñ Ð6Ô7ØôNÐ#ó Nó ó 8ðNðb Ð
‚r)   