o .wi@sxddlmZmZmZmZddlZddlmZddlm Z ddl m Z Gddde Z Gdd d e Z Gd d d e ZdS) )AnyCallableOptionalUnionN)MetricCollection)Metric) WrapperMetriccseZdZdZdeeefdeee fddffdd Z de j de j fd d Z d e j de j fd d Zde j dee j dffddZde j deee fddfddZde fddZde j deee fde fddZdfdd ZZS)MetricInputTransformerzAbstract base class for metric input transformations. Input transformations are characterized by them applying a transformation to the input data of a metric, and then forwarding all calls to the wrapped metric with modifications applied. wrapped_metrickwargsreturnNc s8tjdi|t|ttfstd|||_dS)NzsExpected wrapped metric to be an instance of `torchmetrics.Metric` or `torchmetrics.MetricsCollection`but received )super__init__ isinstancerr TypeErrorr )selfr r  __class__r b/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/torchmetrics/wrappers/transformations.pyrs zMetricInputTransformer.__init__predcC|S)zuDefine transform operations on the prediction data. Overridden by subclasses. Identity by default. r )rrr r rtransform_pred(z%MetricInputTransformer.transform_predtargetcCr)zqDefine transform operations on the target data. Overridden by subclasses. Identity by default. r rrr r rtransform_target0rz'MetricInputTransformer.transform_targetargs.cGsnt|dkr||dfSt|dkr"||d||dfS||d||dg|ddRS)zWWrap transformation functions to dispatch args to their individual transform functions.rN)lenrr)rrr r r_wrap_transform8s  *z&MetricInputTransformer._wrap_transformcOs |j|}|jj|i|dS)z.Wrap the update call of the underlying metric.N)r!r updaterrr r r rr"@s zMetricInputTransformer.updatecCs |jS)z/Wrap the compute call of the underlying metric.)r computerr r rr$Es zMetricInputTransformer.computecOs|j|}|jj|i|S)z/Wrap the forward call of the underlying metric.)r!r forwardr#r r rr&Is zMetricInputTransformer.forwardcs|jtdS)z-Wrap the reset call of the underlying metric.N)r resetrr%rr rr'Ns zMetricInputTransformer.reset)r N)__name__ __module__ __qualname____doc__rrrdictstrrrtorchTensorrrtupler!r"r$r&r' __classcell__r r rrr s*   r c s`eZdZdZ  d dedeeejgejfdeeejgejfde ddf fdd Z Z S) LambdaInputTransformera1Wrapper class for transforming a metrics' inputs given a user-defined lambda function. Args: wrapped_metric: The underlying `Metric` or `MetricCollection`. transform_pred: The function to apply to the predictions before computing the metric. transform_target: The function to apply to the target before computing the metric. Raises: TypeError: If `transform_pred` is not a Callable. TypeError: If `transform_target` is not a Callable. Example: >>> import torch >>> from torchmetrics.classification import BinaryAccuracy >>> from torchmetrics.wrappers import LambdaInputTransformer >>> >>> preds = torch.tensor([0.9, 0.8, 0.7, 0.6, 0.5, 0.6, 0.7, 0.8, 0.5, 0.4]) >>> targets = torch.tensor([1,0,0,0,0,1,1,0,0,0]) >>> >>> metric = LambdaInputTransformer(BinaryAccuracy(), lambda preds: 1 - preds) >>> metric.update(preds, targets) >>> metric.compute() tensor(0.6000) Nr rrr r c shtj|fi||durt|std|d||_|dur2t|s-td|d||_dSdS)NzAExpected `transform_pred` to be of type `Callable` but received ``zCExpected `transform_target` to be of type `Callable` but received `)rrcallablerrr)rr rrr rr rrts  zLambdaInputTransformer.__init__)NN) r(r)r*r+rrrr.r/rrr1r r rrr2Ts"r2c sReZdZdZd deeefdededdffdd Z d e j de j fd d Z Z S)BinaryTargetTransformeraWrapper class for computing a metric on binarized targets. Useful when the given ground-truth targets are continuous, but the metric requires binary targets. Args: wrapped_metric: The underlying `Metric` or `MetricCollection`. threshold: The binarization threshold for the targets. Targets values `t` are cast to binary with `t > threshold`. Raises: TypeError: If `threshold` is not an `int` or `float`. Example: >>> import torch >>> from torchmetrics.retrieval import RetrievalMRR >>> from torchmetrics.wrappers import BinaryTargetTransformer >>> >>> preds = torch.tensor([0.9, 0.8, 0.7, 0.6, 0.5, 0.6, 0.7, 0.8, 0.5, 0.4]) >>> targets = torch.tensor([1,0,0,0,0,2,1,0,0,0]) >>> topics = torch.tensor([0,0,0,0,0,1,1,1,1,1]) >>> >>> metric = BinaryTargetTransformer(RetrievalMRR()) >>> metric.update(preds, targets, indexes=topics) >>> metric.compute() tensor(0.7500) rr thresholdr r Nc s<tj|fi|t|ttfstd|d||_dS)NzBExpected `threshold` to be of type `int` or `float` but received `r3)rrrintfloatrr6)rr r6r rr rrs z BinaryTargetTransformer.__init__rcCs||j|jS)zyCast the target tensor to binary values according to the threshold. Output assumes same type as input. )gtr6todtyperr r rrsz(BinaryTargetTransformer.transform_target)r)r(r)r*r+rrrr8rrr.r/rr1r r rrr5s(r5)typingrrrrr.torchmetrics.collectionsrtorchmetrics.metricrtorchmetrics.wrappers.abstractrr r2r5r r r rs    =5