o yi@sddlmZddlmZmZmZmZddlZddlmZddl m Z ddl m Z ddl mZ dd ed ed efd d ZGddde ZdS)deepcopy)AnyDictOptionalUnionN)Tensor) ModuleList)Metric)apply_to_collectionpoissonsizesampling_strategyreturncCsb|dkrtjd}||f}t|j|ddS|dkr-tjt||dd}|St d) zResample a tensor along its first dimension with replacement. Args: size: number of samples sampling_strategy: the strategy to use for sampling, either ``'poisson'`` or ``'multinomial'`` Returns: resampled tensor r rdim multinomialT) num_samples replacementzUnknown sampling strategy) torch distributionsPoissonsamplearangerepeat_interleavelongrones ValueError)r rpnidxr"W/home/ubuntu/.local/lib/python3.10/site-packages/torchmetrics/wrappers/bootstrapping.py_bootstrap_samplers  r$cseZdZUdZdZeeed<      dded e d ed ed ee e e fd ede deddffdd ZdededdfddZdee e ffddZZS) BootStrapperaUsing `Turn a Metric into a Bootstrapped`_ That can automate the process of getting confidence intervals for metric values. This wrapper class basically keeps multiple copies of the same base metric in memory and whenever ``update`` or ``forward`` is called, all input tensors are resampled (with replacement) along the first dimension. Args: base_metric: base metric class to wrap num_bootstraps: number of copies to make of the base metric for bootstrapping mean: if ``True`` return the mean of the bootstraps std: if ``True`` return the standard diviation of the bootstraps quantile: if given, returns the quantile of the bootstraps. Can only be used with pytorch version 1.6 or higher raw: if ``True``, return all bootstrapped values sampling_strategy: Determines how to produce bootstrapped samplings. Either ``'poisson'`` or ``multinomial``. If ``'possion'`` is chosen, the number of times each sample will be included in the bootstrap will be given by :math:`n\sim Poisson(\lambda=1)`, which approximates the true bootstrap distribution when the number of samples is large. If ``'multinomial'`` is chosen, we will apply true bootstrapping at the batch level to approximate bootstrapping over the hole dataset. kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info. Example:: >>> from pprint import pprint >>> from torchmetrics import BootStrapper >>> from torchmetrics.classification import MulticlassAccuracy >>> _ = torch.manual_seed(123) >>> base_metric = MulticlassAccuracy(num_classes=5, average='micro') >>> bootstrap = BootStrapper(base_metric, num_bootstraps=20) >>> bootstrap.update(torch.randint(5, (20,)), torch.randint(5, (20,))) >>> output = bootstrap.compute() >>> pprint(output) {'mean': tensor(0.2205), 'std': tensor(0.0859)} Tfull_state_update NFr base_metricnum_bootstrapsmeanstdquantilerawrkwargsrc  stjdi|ttstdtfddt|D|_||_||_ ||_ ||_ ||_ d} || vrBtd| d|||_ dS)NzKExpected base metric to be an instance of torchmetrics.Metric but received csg|]}tqSr"r).0_r(r"r# ez)BootStrapper.__init__..)r rz5Expected argument ``sampling_strategy`` to be one of z but recieved r")super__init__ isinstancer rr rangemetricsr)r*r+r,r-r) selfr(r)r*r+r,r-rr.allowed_sampling __class__r1r#r5Ts&  zBootStrapper.__init__argsc Ost|jD]T}t|tt}tt|tt}t|dkr |d}nt|dkr+|d}ntdt||jd |j }t|tt j d|d}t|tt j d|d} |j |j|i| qdS)ztUpdates the state of the base metric. Any tensor passed in will be bootstrapped along dimension 0. rzMNone of the input contained tensors, so could not determine the sampling size)r)rindexN)r7r)r rlenlistrr$rtodevicer index_selectr8update) r9r=r.r! args_sizes kwargs_sizesr sample_idxnew_args new_kwargsr"r"r#rDus     zBootStrapper.updatecCsxtjdd|jDdd}i}|jr|jdd|d<|jr%|jdd|d<|jdur3t||j|d<|jr:||d <|S) zComputes the bootstrapped metric values. Always returns a dict of tensors, which can contain the following keys: ``mean``, ``std``, ``quantile`` and ``raw`` depending on how the class was initialized. cSsg|]}|qSr")compute)r/mr"r"r#r2r3z(BootStrapper.compute..rrr*r+Nr,r-)rstackr8r*r+r,r-)r9 computed_vals output_dictr"r"r#rJs zBootStrapper.compute)r'TTNFr )__name__ __module__ __qualname____doc__r&rbool__annotations__r intrfloatrstrrr5rDrrJ __classcell__r"r"r;r#r%0s< !  !r%)r )copyrtypingrrrrrrtorch.nnr torchmetrics.metricr torchmetrics.utilitiesr rUrWr$r%r"r"r"r#s