Feat: We added the horizon weighing to the distribution losses, simil… #1233
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
|
| @@ -107,100 +107,7 @@ | |||
| "cell_type": "code", | |||
| "execution_count": null, | |||
| "metadata": {}, | |||
| "outputs": [ | |||
There was a problem hiding this comment.
@mwamsojo
If you follow https://github.com/Nixtla/neuralforecast/blob/main/CONTRIBUTING.md and execute
nbdev_clean --clear_all, you will remove this section. According to the guide
If you clean the library's notebooks with this command please backtrack the changes you make to the example notebooks git checkout nbs/examples, unless you intend to change the examples.
But this is part of the document guide that is different from your main work in this PR. So we need to keep this.
There was a problem hiding this comment.
Hello JQGoh, I fixed the issue with the example notebooks in my PR.
…r to point losses like MAE. This has proven useful in our applications. Also cleaning notebook outputs.
mwamsojo
force-pushed
the
Weighted-Horizons-Distribution-Loss
branch
from
bb41743 to
f2c957e
Compare
We suggest modifying the distribution loss class to include the weighting option for distribution losses (to be similar to point losses)