a  dS,@s.ddlmZmZddlZddlmZddlmZmZmZddl m Z eedddZ ee ed d d Z deee ee dd d dZeeeee fdddZdeee ee eedddZeeeee fdddZd eee ee eedddZeeeee fdddZd!eee ee eedddZdS)")OptionalTupleN)Tensor)+_multilabel_confusion_matrix_arg_validation#_multilabel_confusion_matrix_format._multilabel_confusion_matrix_tensor_validation)_cumsum)xreturncCsTt(tj|dddd\}}}Wdn1s60Yt|dd}||S)zRank data based on values.T)sortedZreturn_inverseZ return_countsNrZdim)torchZno_graduniquer)r _inversecountsZranksrw/var/www/html/stable-diffusion-webui/venv/lib/python3.9/site-packages/torchmetrics/functional/classification/ranking.py _rank_datas 6 r)score n_elementsr cCs||S)Nr)rrrrr_ranking_reduce$sr)predstarget num_labels ignore_indexr cCs*t|||||s&td|jdS)NzJExpected preds tensor to be floating point, but received input with dtype )rZis_floating_point ValueErrorZdtype)rrrrrrr%_multilabel_ranking_tensor_validation(sr)rrr cCsnt|}|d||dk<||}|jddd}||dddfkjddtj}||fS)z$Accumulate state for coverage error. rr N)r Z zeros_likeminabssumtofloat32Znumel)rroffsetZ preds_modZ preds_mincoveragerrr!_multilabel_coverage_error_update0s  $r'T)rrrr validate_argsr cCsP|r t|d|dt||||t|||d|dd\}}t||\}}t||S)aCompute multilabel coverage error [1]. The score measure how far we need to go through the ranked scores to cover all true labels. The best value is equal to the average number of labels in the target tensor per sample. Accepts the following input tensors: - ``preds`` (float tensor): ``(N, C, ...)``. Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element. - ``target`` (int tensor): ``(N, C, ...)``. Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). Additional dimension ``...`` will be flattened into the batch dimension. Args: preds: Tensor with predictions target: Tensor with true labels num_labels: Integer specifing the number of labels ignore_index: Specifies a target value that is ignored and does not contribute to the metric calculation validate_args: bool indicating if input arguments and tensors should be validated for correctness. Set to ``False`` for faster computations. Example: >>> from torchmetrics.functional.classification import multilabel_coverage_error >>> _ = torch.manual_seed(42) >>> preds = torch.rand(10, 5) >>> target = torch.randint(2, (10, 5)) >>> multilabel_coverage_error(preds, target, num_labels=5) tensor(3.9000) References: [1] Tsoumakas, G., Katakis, I., & Vlahavas, I. (2010). Mining multi-label data. In Data mining and knowledge discovery handbook (pp. 667-685). Springer US.  thresholdrFr+rZshould_threshold)rrrr'r)rrrrr(r&totalrrrmultilabel_coverage_error:s, r.c Cs| }tjd|jd}|j\}}t|D]p}||dk}t|||}t|dkrt||krt|||} || } n t |} || 7}q(||fS)z5Accumulate state for label ranking average precision.r)devicerr) r tensorr0shaperangerfloatlenmeanZ ones_like) rrZ neg_predsrn_predsn_labelsirelevantZrankingZrankZ score_idxrrr,_multilabel_ranking_average_precision_updateps     r;cCsP|r t|d|dt||||t|||d|dd\}}t||\}}t||S)aCompute label ranking average precision score for multilabel data [1]. The score is the average over each ground truth label assigned to each sample of the ratio of true vs. total labels with lower score. Best score is 1. Accepts the following input tensors: - ``preds`` (float tensor): ``(N, C, ...)``. Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element. - ``target`` (int tensor): ``(N, C, ...)``. Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). Additional dimension ``...`` will be flattened into the batch dimension. Args: preds: Tensor with predictions target: Tensor with true labels num_labels: Integer specifing the number of labels ignore_index: Specifies a target value that is ignored and does not contribute to the metric calculation validate_args: bool indicating if input arguments and tensors should be validated for correctness. Set to ``False`` for faster computations. Example: >>> from torchmetrics.functional.classification import multilabel_ranking_average_precision >>> _ = torch.manual_seed(42) >>> preds = torch.rand(10, 5) >>> target = torch.randint(2, (10, 5)) >>> multilabel_ranking_average_precision(preds, target, num_labels=5) tensor(0.7744) References: [1] Tsoumakas, G., Katakis, I., & Vlahavas, I. (2010). Mining multi-label data. In Data mining and knowledge discovery handbook (pp. 667-685). Springer US. r)r*Fr,)rrrr;r)rrrrr(rrrrr$multilabel_ranking_average_precisions, r<c Cs|j\}}|dk}|jdd}|dk||k@}||}||}||}t|dkrftjd|jddfS|jddjdd}|||tj}d||d} |||} |jdd| | } | |fS)zAccumulate state for label ranking loss. Args: preds: tensor with predictions target: tensor with ground truth labels sample_weight: optional tensor with weight for each sample rr rr)r/g?) r2r"r5r r1r0Zargsortr#r$) rrr7r8r:Z n_relevantmaskrZper_label_lossZ correctionZdenomlossrrr_multilabel_ranking_loss_updates    r?cCsP|r t|d|dt||||t|||d|dd\}}t||\}}t||S)aCompute the label ranking loss for multilabel data [1]. The score is corresponds to the average number of label pairs that are incorrectly ordered given some predictions weighted by the size of the label set and the number of labels not in the label set. The best score is 0. Accepts the following input tensors: - ``preds`` (float tensor): ``(N, C, ...)``. Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element. - ``target`` (int tensor): ``(N, C, ...)``. Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). Additional dimension ``...`` will be flattened into the batch dimension. Args: preds: Tensor with predictions target: Tensor with true labels num_labels: Integer specifing the number of labels ignore_index: Specifies a target value that is ignored and does not contribute to the metric calculation validate_args: bool indicating if input arguments and tensors should be validated for correctness. Set to ``False`` for faster computations. Example: >>> from torchmetrics.functional.classification import multilabel_ranking_loss >>> _ = torch.manual_seed(42) >>> preds = torch.rand(10, 5) >>> target = torch.randint(2, (10, 5)) >>> multilabel_ranking_loss(preds, target, num_labels=5) tensor(0.4167) References: [1] Tsoumakas, G., Katakis, I., & Vlahavas, I. (2010). Mining multi-label data. In Data mining and knowledge discovery handbook (pp. 667-685). Springer US. r)r*Fr,)rrrr?r)rrrrr(r>rrrrmultilabel_ranking_losss, r@)N)NT)NT)NT)typingrrr rZ7torchmetrics.functional.classification.confusion_matrixrrrZtorchmetrics.utilities.datarrintrrr'boolr.r;r<r?r@rrrrsX      6 6#