studio: further enhancements #2510
Conversation
… "experiments" page
There was a problem hiding this comment.
Thanks for documenting the GitHub app stuff @tapadipti! Looks good to me!
I have added two suggestions for things that we could mention in addition.
| @@ -0,0 +1,92 @@ | |||
| # Additional Settings for a View | |||
There was a problem hiding this comment.
@shcheklein This is the page where I think it will be good to lead the users in case of non-dvc-repo errors. This page describes the different ways they can prepare their repo for viewing in Studio, and has a link to the DVC doc. Let me know what you think.
There was a problem hiding this comment.
I think it's complicated still, but it's fine to start here. Later we could use a page similar to this https://dvc-org-studio-docs-v2-kfehbw2.herokuapp.com/doc/user-guide/troubleshooting
And a mechanism with redirects - https://error.dvc.org/lock-issue
| to the sub-directory that contains the DVC repo for which you want to create | ||
| the view. | ||
|
|
||
| - **Data remotes:** If you need to set up DVC data remotes for your view, you |
There was a problem hiding this comment.
@shcheklein Let me know if you think we need to add more details regarding data remotes setup.
There was a problem hiding this comment.
it would be great to mention:
- we dont't read/access data
- we use it only to read plots/metrics files if they are not saved into Git. And people should use it only if they have this situation. They will see a specific message in their View.
There was a problem hiding this comment.
Adding this: "Note that DVC Studio uses the credentials only to read plots/metrics files if they are not saved into Git. It does not access any other data in your remote storage. And you do not need to provide the credentials if any DVC data remote in not used in your Git repository."
casperdcl
changed the title
| <span role="img" aria-label="shield"> | ||
| 🛡 | ||
| <span role="img" aria-label="studio"> | ||
| <img src="img/studio-icon.svg" alt=""></img> |
There was a problem hiding this comment.
not sure if this needs to be className from styled.module.css or even an SVG ReactComponent CC @julieg18 @rogermparent
There was a problem hiding this comment.
Added to #2515 for fixing if needed.
There was a problem hiding this comment.
While it looks like this works totally fine, the first iteration of this alert bar did use a ReactComponent import to display a GitHub icon svg- everything else has been emojis/unicode IIRC.
An SVGR ReactComponent import seems like it would be more optimal, the difference is probably negligible to the point where is ReactComponent causes any problems (like with colors, for example, which didn't come into play on the GH svg), it's fine to stick with img.
There was a problem hiding this comment.
@rogermparent There was some issue when we used img. We were getting Cloudflare alert that the page was offline. It was not breaking all the time, but Ivan noticed it when he was trying to load the page. We replaced the svg img with 🚀 in this PR and the alerts went away (page stopped breaking). Any idea why that may have happened and how we can use images if needed?
There was a problem hiding this comment.
@tapadipti let's please create a proper issue/bug with the screenshot, what we have done to fix, etc
Co-authored-by: Casper da Costa-Luis <work@cdcl.ml>
|
@casperdcl thanks for your review. I've addressed your comments, and created a new ticket with pending issues. #2515 |
Restyle studio: further enhancements
content/docs/studio/get-started.md
Outdated
| The following video provides you a quick overview of DVC Studio. | ||
|
|
||
| https://www.youtube.com/watch?v=hKf4twg832g | ||
|
|
||
| ## What is a view? | ||
|
|
||
| A _view_ is an interactive representation of the experiments run in your ML | ||
| project. DVC Studio automatically identifies datasets, metrics and |
There was a problem hiding this comment.
I think automatically is not the best term here. It doesn't do anything automatically. It parses Git history and specific metafiles (or files you show to it). Automatically sounds like some magic is happening.
| @@ -1,31 +1,38 @@ | |||
| # Get Started with DVC Studio | |||
|
|
|||
| Here, we will walk you through a tutorial to use DVC Studio for collaboration on | |||
| your ML projects. You will need access to a Github, Gitlab or Bitbucket account | |||
| your ML projects. You will need access to a GitHub, GitLab or Bitbucket account | |||
There was a problem hiding this comment.
in the "DVC Studio creates views" view is not defined yet. Better to use some human readable terms - e.g. table with experiments? etc
There was a problem hiding this comment.
Will add a link from 'views' to the 'What is a view?' section on that page. Works?
content/docs/studio/get-started.md
Outdated
| required data (more on this later). Using this data, DVC Studio creates a | ||
| summary view. This includes tables of all the metrics across all your | ||
| experiments. You can also generate plots and compare experiments here. | ||
| hyperparameters in your ML experiments. The data can either be in DVC |
There was a problem hiding this comment.
data has a very specific meaning in ML, here the use of data is misleading
There was a problem hiding this comment.
changing 'The data' to 'These details'
There was a problem hiding this comment.
pls check other places
how about Values ...?
| hyperparameters) are stored in your Git repositories as CSV, JSON or YAML files. | ||
| You can add these details to your Git repositories in two ways: | ||
|
|
||
| 1. **Set up DVC repositories**: You can use [DVC](https://dvc.org/) and Git to |
There was a problem hiding this comment.
DVC is not only about not bloating Git repo. It's about pipelines, connecting then with CML, etc. We should be explaining these benefits better I think
There was a problem hiding this comment.
We should have a separate section/paragraph on DVC and its benefits, and link it to the DVC doc. For now, I'm adding this to the current paragraph: "DVC also
enables you to
share your data and model files,
create data registries,
create data pipelines, connect them with
CML for CI/CD in machine learning, and so on. Find more about the
features and benefits of DVC here."
There was a problem hiding this comment.
ideally we should zoom out and summarize high level benefits here (e.g. pipelines == reproducibility).
One way can be - DVC goes beyond that. It lets you organize your project in a reproducible way, it gives you way to save your experiments holistically as Git commits - data, models, metrics, params ... You could use it to organize manage your data using Git flow (data registries) ...
something like this ...
❗ Please read the guidelines in the Contributing to the Documentation list if you make any substantial changes to the documentation or JS engine.
🐛 Please make sure to mention
Fix #issue(if applicable) in the description of the PR. This causes GitHub to close it automatically when the PR is merged.Please choose to allow us to edit your branch when creating the PR.
Thank you for the contribution - we'll try to review it as soon as possible. 🙏