folivo/PINA
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Solvers logging (#202)

Browse Source 
* Modifying solvers to log every epoch correctly
* add `on_epoch` flag to logger
* fix bug in `pinn.py` `pts -> samples` in `_loss_phys`
* add `optimizer_zero_grad()` in garom generator training loop
* modify imports in `callbacks.py`
* fixing tests

---------

Co-authored-by: Dario Coscia <dariocoscia@dhcp-015.eduroam.sissa.it>
This commit is contained in:
Dario Coscia
2023-11-08 14:10:23 +01:00
committed by Nicola Demo
parent d654259428
commit 3f9305d475
5 changed files with 28 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
pina/solvers/garom.py
View File

@@ -166,6 +166,7 @@ class GAROM(SolverInterface):
Private method to train the generator network.
"""
optimizer = self.optimizer_generator
optimizer.zero_grad()
generated_snapshots = self.generator(parameters)
@@ -258,10 +259,10 @@ class GAROM(SolverInterface):
diff = self._update_weights(d_loss_real, d_loss_fake)
# logging
self.log('mean_loss', float(r_loss), prog_bar=True, logger=True)
self.log('d_loss', float(d_loss), prog_bar=True, logger=True)
self.log('g_loss', float(g_loss), prog_bar=True, logger=True)
self.log('stability_metric', float(d_loss_real + torch.abs(diff)), prog_bar=True, logger=True)
self.log('mean_loss', float(r_loss), prog_bar=True, logger=True, on_epoch=True, on_step=False)
self.log('d_loss', float(d_loss), prog_bar=True, logger=True, on_epoch=True, on_step=False)
self.log('g_loss', float(g_loss), prog_bar=True, logger=True, on_epoch=True, on_step=False)
self.log('stability_metric', float(d_loss_real + torch.abs(diff)), prog_bar=True, logger=True, on_epoch=True, on_step=False)
return

15
pina/solvers/pinn.py
View File

@@ -130,7 +130,7 @@ class PINN(SolverInterface):
if len(batch) == 2:
samples = pts[condition_idx == condition_id]
loss = self._loss_phys(pts, condition.equation)
loss = self._loss_phys(samples, condition.equation)
elif len(batch) == 3:
samples = pts[condition_idx == condition_id]
ground_truth = batch['output'][condition_idx == condition_id]
@@ -138,18 +138,19 @@ class PINN(SolverInterface):
else:
raise ValueError("Batch size not supported")
# TODO for users this us hard to remebeber when creating a new solver, to fix in a smarter way
loss = loss.as_subclass(torch.Tensor)
loss = loss
# add condition losses and accumulate logging for each epoch
condition_losses.append(loss * condition.data_weight)
self.log(condition_name + '_loss', float(loss),
prog_bar=True, logger=True, on_epoch=True, on_step=False)
# TODO Fix the bug, tot_loss is a label tensor without labels
# we need to pass it as a torch tensor to make everything work
# add to tot loss and accumulate logging for each epoch
total_loss = sum(condition_losses)
self.log('mean_loss', float(total_loss / len(condition_losses)),
prog_bar=True, logger=True, on_epoch=True, on_step=False)
self.log('mean_loss', float(total_loss / len(condition_losses)), prog_bar=True, logger=True)
# for condition_loss, loss in zip(condition_names, condition_losses):
# self.log(condition_loss + '_loss', float(loss), prog_bar=True, logger=True)
return total_loss
@property