^j~1\dZddlmZddlmZddlmZddlmZm Z ed eGd d eZ ed eGd d eZ ed eGddeZ ed eGddeZ ed eGddeZgdZy)zSAM2 model configuration)strict)PreTrainedConfig)auto_docstring)CONFIG_MAPPING AutoConfigzfacebook/sam2.1-hiera-tiny) checkpointceZdZUdZdZdZdZeed<dZ eed<dZ eed <d Z ee ezd zed <d Z ee ezd zed <d Zee ezd zed <d Zee ezd zed<d Zee ezd zed<d Ze ed zed<dZeed<d Ze ed zed<d Ze ed zed<d Ze ed zed<d Ze ed zed<d Ze ed zed<dZeed<dZeed<dZeed<dZeed<fdZxZS) Sam2HieraDetConfiga, patch_kernel_size (`list[int]`, *optional*, defaults to `[7, 7]`): The kernel size of the patch. patch_stride (`list[int]`, *optional*, defaults to `[4, 4]`): The stride of the patch. patch_padding (`list[int]`, *optional*, defaults to `[3, 3]`): The padding of the patch. query_stride (`list[int]`, *optional*, defaults to `[2, 2]`): The downsample stride between stages. window_positional_embedding_background_size (`list[int]`, *optional*, defaults to `[7, 7]`): The window size per stage when not using global attention. num_query_pool_stages (`int`, *optional*, defaults to 3): The number of query pool stages. blocks_per_stage (`list[int]`, *optional*, defaults to `[1, 2, 7, 2]`): The number of blocks per stage. embed_dim_per_stage (`list[int]`, *optional*, defaults to `[96, 192, 384, 768]`): The embedding dimension per stage. num_attention_heads_per_stage (`list[int]`, *optional*, defaults to `[1, 2, 4, 8]`): The number of attention heads per stage. window_size_per_stage (`list[int]`, *optional*, defaults to `[8, 4, 14, 7]`): The window size per stage. global_attention_blocks (`list[int]`, *optional*, defaults to `[5, 7, 9]`): The blocks where global attention is used. backbone_configsam2_hiera_det_model` hidden_sizenum_attention_headsr num_channelsN image_sizepatch_kernel_size patch_stride patch_padding query_stride+window_positional_embedding_background_sizenum_query_pool_stagesblocks_per_stageembed_dim_per_stagenum_attention_heads_per_stagewindow_size_per_stageglobal_attention_blocksg@ mlp_ratiogelu hidden_actư>layer_norm_eps{Gz?initializer_rangec |j |jnddg|_|j |jnddg|_|j |jnddg|_|j |jnddg|_|j |jnddg|_|j |j nddg|_|j |j ngd|_|j |jngd|_|j |jngd|_|j |jngd |_ |j |jngd |_ t|0d i|y) Nrr)rrr)r)r)rrr*)r.r*r))r) ) rrrrrrrrrrrsuper __post_init__selfkwargs __class__s v/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/transformers/models/sam2/configuration_sam2.pyr4z Sam2HieraDetConfig.__post_init__Js-1__-H$//tUYl;?;Q;Q;]!7!7deghci151B1B1ND--UVXYTZ373E3E3QT//XY[\W]151B1B1ND--UVXYTZ??K  < <Q 8 :>9N9N9Z 5 5`l(,(@(@(LD $ $Re  372T2T2`D . .fr *+/*D*D*PD & &Vc "-1,H,H,TD ( (Zc $ '') __name__ __module__ __qualname____doc__base_config_key model_typerint__annotations__rrrlistrrrrrrrrrrrr floatr"strr$r&r4 __classcell__r8s@r9r r s]2(O'JK  L#)-Jd3i$&-04sT#Y-4+/L#S /D(/,0M3c?T)0+/L#S /D(/DH/cT1AH!"3")-d3i$&-,0cT)06:!49t#3:.249t+204T#Y-4IuJ NE #u#((r:r ceZdZUdZdZdZdeiZdZe e zdze d<dZ e edze d<dZe dze d<dZee d <d Zee d <d Zee d <d Zee d<dZe edze d<dZee d<dZee d<dZee d<dZee d<fdZxZS)Sam2VisionConfiga backbone_channel_list (`List[int]`, *optional*, defaults to `[768, 384, 192, 96]`): The list of channel dimensions for the backbone. backbone_feature_sizes (`List[List[int]]`, *optional*, defaults to `[[256, 256], [128, 128], [64, 64]]`): The spatial sizes of the feature maps from the backbone. fpn_hidden_size (`int`, *optional*, defaults to 256): The hidden dimension of the FPN. fpn_kernel_size (`int`, *optional*, defaults to 1): The kernel size for the convolutions in the neck. fpn_stride (`int`, *optional*, defaults to 1): The stride for the convolutions in the neck. fpn_padding (`int`, *optional*, defaults to 0): The padding for the convolutions in the neck. fpn_top_down_levels (`List[int]`, *optional*, defaults to `[2, 3]`): The levels for the top-down FPN connections. num_feature_levels (`int`, *optional*, defaults to 3): The number of feature levels from the FPN to use. vision_configsam2_vision_modelr Nbackbone_channel_listbackbone_feature_sizesfpn_hidden_sizerfpn_kernel_size fpn_strider fpn_paddingfpn_top_down_levelsrnum_feature_levelsr!r"r#r$r%r&c |jgdn |j|_|j ddgddgddggn |j|_|jddgn |j|_t|jt rT|jj dd|jd<t|jdd i|j|_n|jt|_t|(d i|y) N)r-r,r+rrN@rrr@rr2) rLrMrS isinstancer dictgetrr r3r4r5s r9r4zSam2VisionConfig.__post_init__s#'#=#=#E 4KeKe "372M2M2Uc3Z#sb"X .[_[v[v #.2-E-E-MAq6SWSkSk d**D 1151E1E1I1I,Xn1oD  .#1$2F2F|2T#U#mX\XlXl#mD  ! ! )#5#7D  ''r:)r;r<r=r>r?r@r sub_configsr rYrrBrLrCrArMrOrPrQrRrSrTr"rEr$rDr&r4rFrGs@r9rIrIes&&O$J:K7;OT,,t3:.249t+2*.D4K.OSOSJK,0cT)0J NE #u#((r:rIceZdZUdZdZdZeed<dZee eze eefzed<dZ ee eze eefzed<dZ eed <d Z eed <d Zeed <dZeed<dZeed<y)Sam2PromptEncoderConfigaY mask_input_channels (`int`, *optional*, defaults to 16): The number of channels to be fed to the `MaskDecoder` module. num_point_embeddings (`int`, *optional*, defaults to 4): The number of point embeddings to be used. scale (`float`, *optional*, defaults to 1): The scale factor for the prompt encoder. prompt_encoder_configrNrr(r patch_sizemask_input_channelsr*num_point_embeddingsr!r"r#r$rscaleN)r;r<r=r>r?rrArBrrCtupler`rarbr"rEr$rDrcr2r:r9r]r]s.OK48Jd3i%S/1846Jd3i%S/16!! !#!J NE E3Nr:r]ceZdZUdZdZdZeed<dZe ed<dZ eed<d Z eed <d Z eed <d Z eed <dZeed<dZeed<dZeed<dZeed<dZeed<dZeed<y)Sam2MaskDecoderConfiga mlp_dim (`int`, *optional*, defaults to 2048): The dimension of the MLP in the two-way transformer. attention_downsample_rate (`int`, *optional*, defaults to 2): The downsample rate for the attention layers. num_multimask_outputs (`int`, *optional*, defaults to 3): The number of multimask outputs. iou_head_depth (`int`, *optional*, defaults to 3): The depth of the IoU head. iou_head_hidden_dim (`int`, *optional*, defaults to 256): The hidden dimension of the IoU head. dynamic_multimask_via_stability (`bool`, *optional*, defaults to `True`): Whether to use dynamic multimask via stability. dynamic_multimask_stability_delta (`float`, *optional*, defaults to 0.05): The stability delta for the dynamic multimask. dynamic_multimask_stability_thresh (`float`, *optional*, defaults to 0.98): The stability threshold for the dynamic multimask. mask_decoder_configrNrr!r"imlp_dimrnum_hidden_layersr.rattention_downsample_raternum_multimask_outputsiou_head_depthiou_head_hidden_dimTdynamic_multimask_via_stabilityg?!dynamic_multimask_stability_deltag\(\?"dynamic_multimask_stability_threshN)r;r<r=r>r?rrArBr"rErhrirrjrkrlrmrnboolrorDrpr2r:r9rfrfs&,OKJGSs  %&s&!"3"NC"",0#T0/3%u304&4r:rfceZdZUdZdZeeedZdZ e e zdze d<dZ e e zdze d<dZe e zdze d<dZee d <fd ZxZS) Sam2Configag prompt_encoder_config (Union[`dict`, `Sam2PromptEncoderConfig`], *optional*): Dictionary of configuration options used to initialize [`Sam2PromptEncoderConfig`]. mask_decoder_config (Union[`dict`, `Sam2MaskDecoderConfig`], *optional*): Dictionary of configuration options used to initialize [`Sam2MaskDecoderConfig`]. Example: ```python >>> from transformers import ( ... Sam2VisionConfig, ... Sam2PromptEncoderConfig, ... Sam2MaskDecoderConfig, ... Sam2Model, ... ) >>> # Initializing a Sam2Config with `"facebook/sam2.1_hiera_tiny"` style configuration >>> configuration = Sam2Config() >>> # Initializing a Sam2Model (with random weights) from the `"facebook/sam2.1_hiera_tiny"` style configuration >>> model = Sam2Model(configuration) >>> # Accessing the model configuration >>> configuration = model.config >>> # We can also initialize a Sam2Config from a Sam2VisionConfig, Sam2PromptEncoderConfig, and Sam2MaskDecoderConfig >>> # Initializing SAM2 vision encoder, memory attention, and memory encoder configurations >>> vision_config = Sam2VisionConfig() >>> prompt_encoder_config = Sam2PromptEncoderConfig() >>> mask_decoder_config = Sam2MaskDecoderConfig() >>> config = Sam2Config(vision_config, prompt_encoder_config, mask_decoder_config) ```sam2)rJr^rgNrJr^rgr%r&c |t|jtrT|jjdd|jd<t |jddi|j|_n|jt d|_t|j trt di|j |_n|j t |_t|jtrtdi|j|_n|jt|_t|(di|y)Nr@rKr2) rXrJrYrZrr^r]rgrfr3r4r5s r9r4zSam2Config.__post_init__s d(($ //3/A/A/E/ElTg/hD  | ,!/0B0B<0P!Q!gTXTfTf!gD     '!/0C!D!FD  d00$ 7)@)^4C]C])^D &  ' ' /)@)BD & d.. 5'<'Xt?W?W'XD $  % % -'<'>D $ ''r:)r;r<r=r>r@r r]rfr[rJrYrrBr^rgr&rDr4rFrGs@r9rsrssx! FJ#!84K 59M4**T18<@4"22T9@:> 0047>#u#((r:rs)rsr rIr]rfN)r>huggingface_hub.dataclassesrconfiguration_utilsrutilsrautorr r rIr]rfrs__all__r2r:r9r{s.3#-78I()I(9I(X786('6(96(r78.9,78!5,!59!5H78A(!A(9A(H r: