torcheval

torcheval copied to clipboard

Minor doc fix: PSNR

2

Summary: correct reference to [`peak_signal_noise_ratio` functional](https://pytorch.org/torcheval/main/generated/torcheval.metrics.functional.peak_signal_noise_ratio.html) in the [`PSNR` class](https://pytorch.org/torcheval/main/generated/torcheval.metrics.PeakSignalNoiseRatio.html) Test plan: Is a documentation fix, does not need tests

plutonium-239

fix: correct reference length calculation

1

### Summary This PR fixes the way brevity penalty (specifically the effective reference corpus length) is calculated in BLEU. Previously, `len_reference` was calculated as `min([len(ref) for ref in references_tokenized])`. However,...

yuxqiu

### 🐛 Describe the bug Attempting to use Torcheval r2_score, but isn't finding the file in Conda environment. I had to copy the file from anaconda3/envs/{myenv}/bin to the local GPU...

123tsnedd

Bug in MulticlassRecall example from when adding one additional class

1

### 🐛 Describe the bug The example from the docs leads to a bug when modified slightly: https://pytorch.org/torcheval/stable/generated/torcheval.metrics.MulticlassRecall.html#torcheval.metrics.MulticlassRecall ``` >>> metric = MulticlassRecall(num_classes=4) >>> input = torch.tensor([[0.9, 0.1, 0, 0],...

jaanli

The FID result cannot be aligned with pytorch-fid/torch-fidelity

1

### 🐛 Describe the bug Just use FrechetInceptionDistance on any images. ```python import torch from torcheval.metrics import FrechetInceptionDistance imgs_1 = ... imgs_2 = ... fid = FrechetInceptionDistance(device=device) fid.update(imgs_1 , True)...

SingleZombie

### 🐛 Describe the bug The score computed by `multiclass_f1_score` for binary classification is wrong. It is not f1 score but accuracy, as shown in following code: ``` import torch...

LittletreeZou

Docs return description of binary_confusion_matrix incorrect

2

Not sure if I'm being daft here but I think this line is incorrect: https://github.com/pytorch/torcheval/blob/e0444c6f7cf89e20b5bd17b16dea4e3793810ba9/torcheval/metrics/functional/classification/confusion_matrix.py#L23 It says that the returned tensor contains the values: ``` ( (true positive, false negative)...

ChrisBog-MV

RetrievalRecall, RetrievalPrecision require different, 1D input than MulticlassRecall, MulticlassPrecision which accept batch input

2

### 🐛 Describe the bug The different behavior of [RetrievalRecall](https://pytorch.org/torcheval/main/generated/torcheval.metrics.RetrievalRecall.html) and [RetrievalPrecision](https://pytorch.org/torcheval/main/generated/torcheval.metrics.RetrievalPrecision.html?highlight=retrievalprecision) make it difficult to compute standard metrics such as Precision@k or Recall@k for multiclass classification problems. Would it...

jaanli

Apply mask for labels for multi-class recall

2

Summary: Fixes https://github.com/pytorch/torcheval/issues/150 Differential Revision: D47241862

ananthsub

Updating `Mean` with 0 leads to 'No calls to update() have been made...' warning

1

### 🐛 Describe the bug When updating a `Mean` metric with a zero, calling `compute()` generates warning that no calls to `update()` have been made. ```python from torcheval import metrics...

dburian