torch

torch copied to clipboard

feature request: available devices as a reflection

3

Is it possible to export a character vector "`available_devices`" that contains the list of supported devices (https://github.com/mlverse/torch/blob/15b780302297a0e4f3cca4303f097c96e590b6e6/src/lantern/src/Device.cpp#L16-L36). If this was available, it is possible for other packages to check that...

sebffischer

Is torch.jit.fork supported inside torch::jit_compile?

1

Thank you for `torch`, switching to GPU powered tensors has dramatically sped up my analyses! I was wondering if torch.jit.fork is supported. I get ``` Error in cpp_jit_compile(source) : attribute...

PhDyellow

I think `names(jit_ops)` should work and return the list of possible operators.

dfalbel

[feature request] Add visualization support of model training

1

Hello, In TensorFlow for R, they implement a visualization tool [tensorboard.R](https://github.com/rstudio/tensorflow/blob/main/R/tensorboard.R), which enables real-time visualization of training process in `viewer` of RStudio. Adding a similar tool in torch for R...

Newtype07

``` # torch_arange(1, t, dtype = "int") # torch_tensor # 1 # 2 # 3 # [ CPUIntType{3} ] # torch_arange(1, t, dtype = "long") # torch_tensor # 1 #...

dfalbel

in pytorch, it is allowed to do `tensor.to(t)` instead of `tensor.to(t.dtype)` ``` >>> a = torch.tensor(1, dtype = torch.int) >>> b = torch.tensor(1, dtype = torch.float) >>> b.to(a) tensor(1, dtype=torch.int32)...

Athospd

deadlock when training torch model

8

Hi @dfalbel , I'm having an issue that the model training just hangs when training models. This doesn't happen always but has now happened twice in the last week. I'm...

egillax

This issue aggregates some larger features that would be great additions to the torch package. - [ ] #738 - [ ] #928 - [x] #929 (https://github.com/mlverse/torch/pull/1020, https://github.com/mlverse/torch/pull/1023) - [...

dfalbel

cannot open URL 'https://storage.googleapis.com/torch-datasets/mnist/train-images-idx3-ubyte.gz

2

Hello @dfalbel, recalling #1006, this new issue originates from the same problem that the firewall is blocking access to [https://storage.googleapis.com/torch-lantern-builds/binaries/668fb380a07a2dfb90b57a28b3a07af11c9188ff/lantern-0.9.1.9001+cu116-win64.zip](url). I run the following code, as mentioned in **Section 13.2.3**...

Milardkh

model to ONNX

4

Is it possible to add a model converter to ONNX ?

VladPerervenko