-
Notifications
You must be signed in to change notification settings - Fork 52
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add changelog #28
Add changelog #28
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A few tiny things, but otherwise this is great. Let's try to merge this first, so we can add the changes from other PRs to the changelog.
CHANGELOG.md
Outdated
|
||
## [v0.1.0](https://github.com/joeloskarsson/neural-lam/releases/tag/v0.1.0) | ||
|
||
First tagged release of `neural-lam`, matching Oscarsson et al 2023 publication |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
* Oskarsson :)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh no, sorry! I keep doing this... so sorry
CHANGELOG.md
Outdated
[c14b6b4](https://github.com/joeloskarsson/neural-lam/commit/c14b6b4323e6b56f1f18632b6ca8b0d65c3ce36a) | ||
@joeloskarsson | ||
|
||
- compute `rmse` after spatial averaging |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The change is to take sqrt after averaging all the forecasts. It was always after spatial averaging. See e.g. difference between v1 and v2 of WB2 https://arxiv.org/abs/2308.15560.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't quite understand what you mean by "averaging all the forecasts". Is there a figure in the WeatherBench2 article you're referring? Or is it something discussed in v2 that was missing in v1 of that article?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry, I realize that was not very clear. I was referencing the RMSE definition in the WB2 paper (eq.2 in both versions). It is somewhat different in v1 and v2. A good description for this change would be "change RMSE definition to compute sqrt after all averaging".
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
thank you :)
CHANGELOG.md
Outdated
|
||
### Removed | ||
|
||
- `WeatherDataset(torch.Dataset)` no longer returns "static" component of |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe better to refer to this as batch-static here as well, to avoid confusing it with actual static features, which were never returned from WeatherDataset
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes, good point, I'll change that
thanks for your feedback @joeloskarsson! I've adapted the changelog with your suggestions. @sadamov are you happy with the changelog as i stands? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All looks good now
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for adding the CHANGELOG, looks good! Ready to merge from my side.
Do we agree to squash commits per default?
Go ahead and merge. Yes, we should definitely squash as a default. If there is a setting to to strictly enforce this that would be good. |
This PR adds a changelog as I proposed in #7. I have tried to go through the commits (v0.1.0...HEAD) since v0.1.0
@joeloskarsson and @sadamov would you able to check this with the commits/PRs that you are tagged on to check that you are happy with what I've written? Otherwise maybe you could suggest some changes?
You can view the rendered changelog here: https://github.com/leifdenby/neural-lam/blob/maint/changelog/CHANGELOG.md