Dave Berenbaum
Dave Berenbaum
See https://github.com/iterative/example-repos-dev/blob/b8a8f0af39b88c735af0b8065c37b609242b53ed/example-get-started/code/src/evaluate.py#L61. I have to keep two separate live instances to be able to log sklearn plots for both train and test data because there's no way to create multiple...
Follow up to #310 > Before creating the experiment ref, DVCLive should write to a `dvc.yaml` file inside the `Live` directory, which should configure parameters, metrics, and plots logged by...
Depends on https://github.com/iterative/dvc/issues/6746. If DVC is installed, DVCLive should be able to create an experiment ref. This should be possible by either: 1. Making `Live` a context manager invoked by...
Render the report inline within a notebook instead of in a separate HTML/mardown document. Should be prioritized after we have a basic DVCLive-only workflow for parameters/metrics/plots.
When inside a DVC repo, DVCLive should be able to automatically cache large outputs like plots and models without asking users to configure them in pipeline stage outputs.
@shcheklein in that case our concurrency level will be `jobs * jobs`, which is generally going to be way too high in the default case. I also considered splitting `jobs`...
Follow up to https://github.com/iterative/dvc/issues/7754. It's common to have essentially a one-column csv with predicted or actual values for your data. See https://dvc.org/doc/command-reference/plots/show#sourcing-x-and-y-from-different-files for an example. However, it's unlikely that these...
Templating and foreach are only supported in `stages` in `dvc.yaml`. It would be nice to have support in top-level `params`, `metrics`, `plots`, and `artifacts`.
We have `dvc artifacts get`, but it would be useful to also have `dvc artifacts import` to enable using artifacts in downstream dvc projects. More info in [slack](https://iterativeai.slack.com/archives/CUSNDR35K/p1701442649647219?thread_ts=1701378725.620539&cid=CUSNDR35K).
## Summary / Background Provide robust logging for experiment runs. ## Scope When running any experiment, save logs of the output, errors, hardware usage, time ,etc. Be able to retrieve...