.`iUdZddlmZddlmZddlmZejeZ ddddd Z e e e eeeeffed <d Zd ZGd deZdS)z Radio vision model configuration)Any)PretrainedConfig)logging)i i)irri )ii)i r i)vit_small_patch16_224vit_base_patch16_224vit_large_patch16_224vit_huge_patch16_224VIT_TIMM_DIM_BY_NAME)g3<4'?gwg M?gy{ ?)gB91?gwt.?g U?c eZdZdZdZddddddd d d eed d dfd ededede de dede de dedede e e e fe zde e e e fe zded zde e eefd zde ffd ZxZS) RadioConfiga> This is the configuration class to store the configuration of a Radio vision model. It is used to instantiate a Radio model according to the specified arguments, defining the model architecture. Args: model_name: Name of the vision transformer model (e.g., "vit_base_patch16_224"). Used to determine architecture dimensions from `VIT_TIMM_DIM_BY_NAME`. image_size: The size (resolution) of each image. patch_size: The size (resolution) of each patch. qkv_bias: Whether to add a bias to the queries, keys and values. qk_normalization: Whether to apply normalization to queries and keys. norm_type: The normalization type to use. layer_norm_eps: The epsilon used by the layer normalization layers. initializer_factor: A factor for initializing all weight matrices. hidden_act: The non-linear activation function in the encoder. cpe_max_size: Maximum image size for position embeddings. norm_mean: Mean values for image normalization (RGB channels). Defaults to (0.48145466, 0.4578275, 0.40821073)). norm_std: Standard deviation values for image normalization (RGB channels). Defaults to (0.26862954, 0.26130258, 0.27577711)). register_multiple: Number of register tokens to use. teachers: A list of teacher model configurations. Each teacher configuration is a dict with keys like "name" and some may have "use_summary". cls_token_per_teacher: Whether to use a separate CLS token for each teacher. radior TF layer_normgư>g?geluiN model_name image_size patch_sizeqkv_biasqk_normalization norm_typelayer_norm_epsinitializer_factor hidden_act cpe_max_size norm_meannorm_stdregister_multipleteacherscls_token_per_teacherc ||_t|\|_|_|_|_||_||_||_||_ ||_ ||_ ||_ | |_ | |_t| t t"frt#| n| |_t| t t"frt#| n| |_| |_||ng|_||_t/jdi|dS)N)rr hidden_sizenum_hidden_layersnum_attention_headsintermediate_sizerrrrrrrrr isinstancetuplelistr r!r"r#r$super__init__)selfrrrrrrrrrrr r!r"r#r$kwargs __class__s y/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/vllm/transformers_utils/configs/radio.pyr/zRadioConfig.__init__6s &% ! ,     "  $  "$$  0","4$())eT]CC RDOOO )E4=AA ODNNNx "3$,$8b %:"""6""""")__name__ __module__ __qualname____doc__ model_typeOPENAI_CLIP_MEANOPENAI_CLIP_STDstrintboolfloatr,r-dictrr/ __classcell__)r2s@r3rrsK8J !&% $$'  7G6E(,04&+!,#,#,#,# ,#  ,#  ,#,#,#",#,#,#u,-4,#ue+,t3,#:,#tCH~&-,# $!,#,#,#,#,#,#,#,#,#,#r4rN)r8typingr transformers.configuration_utilsrtransformers.utilsr get_loggerr5loggerrr@r<r,r=__annotations__r:r;rr&r4r3rHs'&&======&&&&&&  H % %0/10 >>d3c3S&8 99:76K#K#K#K#K#"K#K#K#K#K#r4