numpy like tensor.all and tensor.any #2481
Comments
|
As a note, |
|
I didn't know. Thanks! |
mlaradji
added a commit
to mlaradji/LCFCN
that referenced
this issue
Dec 26, 2018
This is a proposed change from `np.all()` to `torch.tensor.all()` in `datasets/trancos.py`. I don't know if it is necessary, but it seems like it shouldn't break anything, and it does get rid of the following error for me:
```
>> python main.py -m train -e trancos -r
Model: ResFCN - Dataset: trancos - Metric: MAE
Starting from scratch...
Training Epoch 1 .... 403 batches
Traceback (most recent call last):
File "main.py", line 45, in <module>
main()
File "main.py", line 36, in main
train.train(dataset_name, model_name, metric_name, path_history, path_model, path_opt, path_best_model, args.reset)
File "/home/mlaradji/projects/ElementAI/LCFCN/train.py", line 76, in train
epoch=epoch)
File "/home/mlaradji/projects/ElementAI/LCFCN/utils.py", line 28, in fit
for i, batch in enumerate(dataloader):
File "/home/mlaradji/.conda/envs/LCFCN/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 615, in __next__
batch = self.collate_fn([self.dataset[i] for i in indices])
File "/home/mlaradji/.conda/envs/LCFCN/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 615, in <listcomp>
batch = self.collate_fn([self.dataset[i] for i in indices])
File "/home/mlaradji/projects/ElementAI/LCFCN/datasets/trancos.py", line 63, in __getitem__
if np.all(points == -1):
File "/home/mlaradji/.local/lib/python3.6/site-packages/numpy/core/fromnumeric.py", line 2089, in all
return _wrapreduction(a, np.logical_and, 'all', axis, None, out, keepdims=keepdims)
File "/home/mlaradji/.local/lib/python3.6/site-packages/numpy/core/fromnumeric.py", line 81, in _wrapreduction
return reduction(axis=axis, out=out, **passkwargs)
TypeError: all() missing 1 required positional arguments: "dim"
```
According to this[pytorch/pytorch#2481], `torch.tensor.all` is apparently not well-documented but it exists.
samnordmann
pushed a commit
to samnordmann/pytorch
that referenced
this issue
Mar 6, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi all!
As mentioned in issue #2228, it would be nice to have functions close to the numpy API. Currently, logic functions like np.all and np.any are still missing. Of course, workarounds can be easily achieved, but having such a function would increase readability.
A simple implementation could consist of using boolean_tensor.min() != 0 and boolean_tensor.max() == 1.
The text was updated successfully, but these errors were encountered: