David de la Iglesia Castro
David de la Iglesia Castro
Depending on the type of model to be trained, the time in between calls to `next_step` may vary significantly. In common deep learning scenarios, i.e. the [keras callback](https://github.com/iterative/dvclive/blob/master/dvclive/keras.py), `next_step` is...
https://github.com/iterative/vscode-dvc/pull/1673#discussion_r866510987
can we rename it? is it a default name? it's a bit confusing to have `dvclive` dir, `dvclive/scalars` which are actually plots and `dvclive.json` that is actually scalars ... how...
I think having this here also leads to the old "experiment is logged with the last epoch number" error _Originally posted by @mattseddon in https://github.com/iterative/vscode-dvc/pull/1618#discussion_r861417166_
> It might be enough to hint about the environment variable like in https://github.com/iterative/dvc/blob/f84f9cf1846554a13f388693220378edb8d17be6/dvc/commands/plots.py#L111. @daavoo What do you think about this idea? _Originally posted by @dberenbaum in https://github.com/iterative/dvclive/pull/243#discussion_r859756930_
I think we should keep the path in envvar in an absolute form in `DVCLIVE_PATH`. Otherwise, the user's script could change the working directory similar to following: ```suggestion (tmp_dir /...
Reinforcement learning algorithms also use the concept of `steps` or iterations so the support could be similar to the existing integrations with Deep Learning frameworks. There are a lot of...
`log_image` currently supports `numpy` and `PIL` as inputs. It would be nice to also support `matplotlib` figures
@daavoo I think that this is quite a challenge for us to decide. 1. On one hand we don't want to restrict the dependencies, because that could interfere with users...
The latest version breaks current integration https://github.com/catalyst-team/catalyst/releases/tag/v22.02 Failure: https://github.com/iterative/dvclive/runs/5183975198?check_suite_focus=true#step:6:156