Piotr Dabkowski

Results 9 issues of Piotr Dabkowski

### Current Behavior Long audio captions are clipped and there is no way to see them in full (likely also applies to other media, but did not test). They are...

app

### Current Behavior Wandb is prunning data points from graphs, I guess to save memory and improve performance. Unfortunately, this seems to happen also for Validation charts, where the number...

app

### Current Behavior A large fraction of the logged media is missing (audio & images), refreshing page sometimes help, the issue seems to be transient and comes and goes. This...

app

Thanks a lot for this project. We would like to try our model on this dataset and compare to results by other teams. Is this dataset (AESRC2020) available anywhere? The...

First of all, thank you for the project, I think it is really useful, especially that the official NVIDIA implementation is not released yet! Did you manage to train the...

I have noticed that alphas is not a ModuleList, but rather a list of parameters, this causes them to be not trainable. Example: [self.alpha1 = [nn.Parameter(torch.ones(1, channels, 1).to(rank)) for i...

Fixes #8378. Audio from the response should be used in the audio element, rather than using potentially incomplete URL as the src. This is similar to how swagger-ui does it...

Activation checkpointing recomputes the activations and hence it will need to re-execute parts of forward pass. This re-execution should not affect history and be allowed. Currently this error is being...

### Bug description When `val_check_interval` is used, and the model is resumed the validation is run immediately after resume, wasting resources (causing ckpt save, and whole validation run). https://github.com/Lightning-AI/pytorch-lightning/discussions/18110 ###...

bug
needs triage
ver: 2.4.x