tjddlZddlmZmZddlmZddlZddlZddl m cm Z ddlm Z m Z ddlmZddlmZddlmZmZdd lmZdd lmZdd lmZmZmZmZmZdd l m!Z!m"Z"dd l#m$Z$ddl%m&Z&ddl'm(Z(m)Z)ddl*m+Z+m,Z,m-Z-ddl.m/Z/ddl0m1Z1e)deGddeZ2e)deGddeZ3e)deGddeZ4e)deGdd eZ5e)d!eGd"d#eZ6Gd$d%e j7Z8e&d&'d(e9d)e9d*ej:d+ej;d,ej f d-Z< dd.ej d/e=dzd0e=dzd1e=dzd,ej f d2Z>Gd3d4e j7Z?ed5Gd6d7e j7Z@d8ZA dd:e j7d;ej dej dzd?e=e9zd@e=dzdAe=dzd,eBej ej ffdBZCdCej dDej dEej dFej d,eBej ej ff dGZDdHej dIe9d,ej fdJZEGdKdLe j7ZFGdMdNe j7ZGGdOdPe j7ZHGdQdRe j7ZIGdSdTe j7ZJGdUdVeZKGdWdXe j7ZLGdYdZe j7ZMGd[d\e j7ZNGd]d^e j7ZOGd_d`e j7ZPe)Gdadbe"ZQGdcddeQZRe)GdedfeQZSe)GdgdheeQZTe)dijGdkdleQZUddnZVe)dodpqGdrdseQZWe)dtduqGdvdweQZXe)dxdyqGdzd{eQZYe)d|d}qGd~deQZZgdZ[dS)N)CallableIterable) dataclass)Tensornn)initialization)ACT2FN) BackboneMixinfilter_output_hidden_states)use_kernel_forward_from_hub)GradientCheckpointingLayer)BackboneOutputBaseModelOutputBaseModelOutputWithPooling ModelOutputSemanticSegmenterOutput)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)#compile_compatible_method_lru_cache)TransformersKwargsauto_docstring)can_return_tuplemaybe_autocastmerge_with_config_defaults)capture_outputs)Sapiens2Configz Output type of [`Sapiens2Backbone`], extending [`BackboneOutput`] with optional CLS tokens from each selected feature stage (used when `config.return_class_token=True`). ) custom_introc>eZdZUdZdZeejdzed<dS)Sapiens2BackboneOutputz cls_tokens (`tuple(torch.FloatTensor)`, *optional*): CLS token from each selected feature stage, each of shape `(batch_size, hidden_size)`. Only present when `config.return_class_token=True`. N cls_tokens) __name__ __module__ __qualname____doc__r#tupletorch FloatTensor__annotations__l/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/transformers/models/sapiens2/modeling_sapiens2.pyr"r"1s; 37Je'(4/66666r-r"z6 Class for outputs of pose estimation models. ceZdZUdZdZejdzed<dZejdzed<dZ e ejdfdzed<dZ e ejdfdzed<dS)Sapiens2PoseEstimatorOutputa loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Pose estimation loss. heatmaps (`torch.FloatTensor` of shape `(batch_size, num_keypoints, height, width)`): Heatmaps as predicted by the model. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each stage) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states (also called feature maps) of the model at the output of each stage. Nlossheatmaps. hidden_states attentions) r$r%r&r'r1r)r*r+r2r3r(r4r,r-r.r0r0Es  &*D% d "))))-He$&---:>M5*C/047>>>7;Je',-4;;;;;r-r0z8 Class for outputs of normal estimation models. ceZdZUdZdZejdzed<dZejdzed<dZ e ejdfdzed<dZ e ejdfdzed<dS)Sapiens2NormalEstimatorOutputa loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Normal estimation loss. normals (`torch.FloatTensor` of shape `(batch_size, num_labels, height, width)`): Raw normal map predictions as output by the model (unnormalized). hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each stage) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the initial embedding outputs. attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`): Tuple of `torch.FloatTensor` (one per layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights after the attention softmax. Nr1normals.r3r4) r$r%r&r'r1r)r*r+r7r3r(r4r,r-r.r6r6]s  &*D% d ")))(,GU  %,,,:>M5*C/047>>>7;Je',-4;;;;;r-r6z: Class for outputs of pointmap estimation models. ceZdZUdZdZejdzed<dZejdzed<dZ ejdzed<dZ e ejdfdzed<dZ e ejdfdzed<dS) Sapiens2PointmapEstimatorOutputa loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Pointmap estimation loss. pointmaps (`torch.FloatTensor` of shape `(batch_size, 3, height, width)`): Per-pixel 3D XYZ coordinate predictions in canonical camera space. scales (`torch.FloatTensor` of shape `(batch_size, 1)`, *optional*): Canonical focal length / actual focal length ratio. `None` when no scale branch is configured. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each stage) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the initial embedding outputs. attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`): Tuple of `torch.FloatTensor` (one per layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights after the attention softmax. Nr1 pointmapsscales.r3r4) r$r%r&r'r1r)r*r+r:r;r3r(r4r,r-r.r9r9xs &*D% d ")))*.Iu 4'...'+FE  $+++:>M5*C/047>>>7;Je',-4;;;;;r-r9z4 Class for outputs of image matting models. ceZdZUdZdZejdzed<dZejdzed<dZ e ejdzed<dZ e ejdzed<dZ ejdzed<dS)Sapiens2ImageMattingOutputaN loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Loss. alphas (`torch.FloatTensor` of shape `(batch_size, 1, height, width)`): Estimated alpha values. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each stage) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states (also called feature maps) of the model at the output of each stage. foregrounds (`torch.FloatTensor` of shape `(batch_size, 3, height, width)`): Pre-multiplied RGB foreground predictions in `[0, 1]` (sigmoid-activated). Nr1alphasr3r4 foregrounds) r$r%r&r'r1r)r*r+r>r3r(r4r?r,r-r.r=r=s  &*D% d ")))'+FE  $+++59M5*+d299926Je'(4/666,0K"T)00000r-r=cbeZdZdZdeffd Zd dejdejdzdejfdZxZ S) Sapiens2EmbeddingszM Construct the CLS token, mask token, position and patch embeddings. configct||_tjt jdd|j|_|j r-tjt j dd|jnd|_ tjt j d|j |j|_tj|j|j|j|j|_dS)Nr) kernel_sizestride)super__init__rBr Parameterr)randn hidden_size cls_tokenuse_mask_tokenzeros mask_tokenemptynum_register_tokensregister_tokensConv2d num_channels patch_sizepatch_embeddingsselfrB __class__s r.rGzSapiens2Embeddings.__init__s  ek!Q8J&K&KLLQWQfp",u{1a9K'L'LMMMlp!|EK6;UW]Wi,j,jkk "  !3AR[a[l! ! ! r-N pixel_valuesbool_masked_posreturncL||jtd|jd}|jjj}|||}|ddd}|H|j|j}tj | d||}|j |dd}|j |dd}tj|||gd} | S)Nz:bool_masked_pos requires use_mask_token=True in the configrdtyperdim)rN ValueErrorshaperUweightr^toflatten transposer)where unsqueezerKexpandrQcat) rWrYrZ batch_size target_dtyperUrNrKrQ embeddingss r.forwardzSapiens2Embeddings.forwards  &4?+BYZZ Z!'* ,39  00|1T1TUU+33A66@@AFF  &++,<,BCCJ${?+D+DR+H+H*Vfgg N))*b"== .55j"bIIY ?ZZF^^Aq))F[[^^F)F##C)F##C+ $ $ $ $ $ $ $ $ $ $ $ $ $ $ $."vvEv""CFFF$7$777s1CEEE) r$r%r&r)rr+rrGr(rprrrss@r.rrsl5~555555" 8EL 8U5<;U5V 8 8 8 8 8 8 8 8r-rRMSNormcTeZdZd deddffd ZdejdejfdZdZxZ S) Sapiens2RMSNormư>epsr[Ncttjt j||_||_dS)z> Sapiens2RMSNorm is equivalent to T5LayerNorm N)rFrGrrHr)onesrevariance_epsilon)rWrJrrXs r.rGzSapiens2RMSNorm.__init__JsD l5:k#:#:;; #r-r3c|j}|tj}|ddd}|tj||jzz}|j||zS)Nr_r`T)keepdim) r^rfr)rpowmeanrsqrtrre)rWr3 input_dtypevariances r.rpzSapiens2RMSNorm.forwardRs|#) %((77  $$Q'',,R,>>% Ht?T4T(U(UU {]--k::::r-cHt|jjd|jS)Nz, eps=)r(rerdrrWs r. extra_reprzSapiens2RMSNorm.extra_reprYs& )**II$2GIIIr-)r) r$r%r&floatrGr)rrprrrrss@r.rrHs$$$$$$$$$$;U\;el;;;;JJJJJJJr-rc|dd|jddzf}|d|jddzdf}tj| |fdS)z*Rotates half the hidden dims of the input..Nr`r_ra)rdr)rl)xx1x2s r. rotate_halfr]s] 3"!'"+"" " #B 3 q "" " #B 9rc2YB ' ' ''r-modulequerykeyvalueattention_maskdropoutscalingsoftcapc l| |jdz}t||j} t||j} tj|| dd|z} || |z } tj| } | |z} || |z} tj | dtj  |j } tj | ||j} tj| | } | dd} | | fS)Nr_rr`)rbr^)prr)r repeat_kvnum_key_value_groupsr)matmulrhtanhr functionalsoftmaxrrfr^rr contiguous) rrrrrrrrkwargs key_states value_states attn_weights attn_outputs r.eager_attention_forwardrds%/4'3 ;<