From 10d1d6e3417c2d93ac19bd65b44a1a1b92a37bb6 Mon Sep 17 00:00:00 2001 From: Thomas M Kehrenberg Date: Fri, 6 Oct 2023 18:28:33 +0200 Subject: [PATCH] Fix two small mistakes --- external_confs/experiment/nicopp/rn18/only_pred_y_loss.yaml | 2 +- src/algs/adv/supmatch.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/external_confs/experiment/nicopp/rn18/only_pred_y_loss.yaml b/external_confs/experiment/nicopp/rn18/only_pred_y_loss.yaml index c7721ead..4ea2b90a 100644 --- a/external_confs/experiment/nicopp/rn18/only_pred_y_loss.yaml +++ b/external_confs/experiment/nicopp/rn18/only_pred_y_loss.yaml @@ -22,7 +22,7 @@ alg: steps: 30000 use_amp: true pred: - lr: ${ ae lr } + lr: ${ ae.lr } log_freq: 100000000000 # never val_freq: 1000 diff --git a/src/algs/adv/supmatch.py b/src/algs/adv/supmatch.py index e01c623a..7768d989 100644 --- a/src/algs/adv/supmatch.py +++ b/src/algs/adv/supmatch.py @@ -28,7 +28,7 @@ def _get_data_iterators(self, dm: DataModule) -> tuple[IterTr, IterDep]: if dm.deployment_ids is None: logger.warning( "Support matching is enabled but without any balancing of the deployment set " - "- this can be achieved either by setting 'deployment_ids'." + "- this can be achieved by setting 'deployment_ids'." ) dl_tr = dm.train_dataloader(balance=True) # The batch size needs to be consistent for the aggregation layer in the setwise neural