SolomidHero

LightweightConv. Incorrect decoding with incremental state

1

## 🐛 Bug I am using LightweightConv layer and it seems to be incorrect when applying with incremental state and step more than 1: https://github.com/pytorch/fairseq/blob/fcca32258c8e8bcc9f9890bf4714fa2f96b6b3e1/fairseq/modules/lightweight_convolution.py#L221 ### Code and Expected behavior...

transformations with sample rate bug

6

Hi! In current implementation after resample transformation the current sample rate could be changed and not given to the future state of composition pipeline. Thus, such code may lead to...

Running hydra configured scripts

4

**Is your feature request related to a problem? Please describe.** It is impossible to run profiling on hydra (https://github.com/facebookresearch/hydra) configured python scripts. Hydra scripts often runs in such manner: ```...

### **Describe the bug** Can't run profiling of script if that script is a submodule which uses relative parent imports. ### **To Reproduce** Assume next structure: ``` A/ __init__.py payload.py...

filtfilt consistency with scipy

5

### 🐛 Describe the bug Hello! I tried using created in #1681 `torchaudio.functional.filtfilt` instead of `scipy.signal.filtfilt`. In my observation there is no such combination of arguments that filtered output is...

Loading finetuned checkpoint. Local method not working.

10

Hi! For some time I though it was a correct way to load pretrained named model (such as `tera`, `mockingjay`, `cpc`) provided in `example_extract.py` files: ```python ckpt_path=`/path/to/valid/ckpt-1337steps.pt` Upstream_local = getattr(importlib.import_module('hubconf'),...

.f0 and .pow features

2

Hi! I am training the model for my custom dataset and there is a problem with .f0 and .pow in `egs/cyclevae/conf` folder. As I understand it right: - `spk_name.f0` contains...

cannot import name 'search_models' from 'sparsezoo'

2

**Describe the bug** can't import search_models from sparsezoo **Expected behavior** ```py3 from sparsezoo import search_models ``` runs ok **Environment** Google colab environment with gpu. ``` sparsezoo==1.0.0 torch @ https://download.pytorch.org/whl/cu113/torch-1.12.0%2Bcu113-cp37-cp37m-linux_x86_64.whl torchaudio...

Hi! I experienced some troubles with inference based on model inner operations: It seems like for pytorch 1.7.0, `torch.nn.utils.rnn.pack_padded_sequence`'s src_length must be in cpu, even if we're using cuda: pytorch/pytorch#43227...

torch.full can't use implicit scalars

1

**Describe the issue**: For some functions like `torch.full` it is needed to have input from other tensors. For example I use it to create lengths array, when it is not...