Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #678 +/- ##
=======================================
Coverage 87.35% 87.35%
=======================================
Files 82 82
Lines 8476 8476
Branches 581 581
=======================================
Hits 7404 7404
Misses 922 922
Partials 150 150 🚀 New features to boost your workflow:
|
|
i like this a lot! thanks can you make sure the linter is happy? |
|
View / edit / reply to this conversation on ReviewNB marcocuturi commented on 2026-03-15T10:27:23Z use the citation format that we use in ott-jax and add citations to |
|
View / edit / reply to this conversation on ReviewNB marcocuturi commented on 2026-03-15T10:27:23Z Line #4. class MLP(nn.Module): isn't this model class already implemented within the toolbox? |
|
View / edit / reply to this conversation on ReviewNB marcocuturi commented on 2026-03-15T10:27:24Z Line #28. def step(v_params, key: jax.Array, x0: jax.Array, x1: jax.Array): same, shouldn't we use the toolbox implementation? if something is missing, can we consider changing it there? |
|
View / edit / reply to this conversation on ReviewNB marcocuturi commented on 2026-03-15T10:27:25Z Line #15. def solve_semidiscrete( can't we jit the solver and do away with this helper function? since the logic is only instantiating the geom + using Semidiscretesolver. @michalk8 might have a different opinion zsteve commented on 2026-03-19T00:36:12Z agreed! |
|
View / edit / reply to this conversation on ReviewNB marcocuturi commented on 2026-03-15T10:27:25Z cool :) |
|
View / edit / reply to this conversation on ReviewNB marcocuturi commented on 2026-03-15T10:27:26Z Line #2. def _pair(x0: jax.Array, x1: jax.Array, aux, what: str): can we consider having three zsteve commented on 2026-03-19T00:36:36Z I've switched to setting up different dataloaders for this instead. |
|
agreed! View entire conversation on ReviewNB |
|
I've switched to setting up different dataloaders for this instead. View entire conversation on ReviewNB |
2 participants
I'm putting together a simple demo, this is similar to the existing OTFM tutorial but perhaps shows the distinction between IFM/OTFM/SDFM more clearly. I've realized that some of the basic functionalities (vector field, dataloaders, etc.) are also implemented as part of OTT so need to fix that at some point.
download.mp4