All the plots are now dvclive, which simplifies the code and dvc.yaml a bit. I ended up caching all of them, because I don't see much upside to keeping the tsv/json plots in git, and it simplifies dvc.yaml. We lose explaining how to make more custom plots, but I think that's outside the scope of get started. If we want more custom plots, we can try to think of more complex visualizations to include.
Todo for deployment:
[x] Deploy the updated code
[x] Deploy the updated repo (experiments, PRs, etc.; see readme)
Closes https://github.com/iterative/example-repos-dev/issues/234.
Updates two plots to use dvclive:
log_sklearn_plot
with additional kwargdrop_intermediate=True
)log_image
matplotlib support)All the plots are now dvclive, which simplifies the code and
dvc.yaml
a bit. I ended up caching all of them, because I don't see much upside to keeping the tsv/json plots in git, and it simplifiesdvc.yaml
. We lose explaining how to make more custom plots, but I think that's outside the scope of get started. If we want more custom plots, we can try to think of more complex visualizations to include.Todo for deployment: