Bump pytorch lightning to 1.1.7

benchmarks/common.py

@@ -17,8 +17,8 @@ def setup(train_path, fixed_input_height=0):
 
     syms = str(train_path / "syms")
     syms_table = SymbolsTable()
-    for k, v in data_module.syms.items():
-        syms_table.add(v, k)
+    for k, v in data_module.syms:
+        syms_table.add(k, v)
     syms_table.save(syms)
 
     model(

laia/callbacks/learning_rate.py

@@ -1,12 +1,13 @@
 import pytorch_lightning as pl
+from pytorch_lightning.callbacks.lr_monitor import LearningRateMonitor
 from pytorch_lightning.utilities import rank_zero_only
 
 import laia.common.logging as log
 
 _logger = log.get_logger(__name__)
 
 
-class LearningRate(pl.callbacks.LearningRateMonitor):
+class LearningRate(LearningRateMonitor):
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
         self.last_values = None
@@ -23,8 +24,8 @@ def on_train_start(self, trainer, *args, **kwargs):
         self.last_values = {}
 
     @rank_zero_only
-    def on_epoch_end(self, trainer, *args, **kwargs):
-        super().on_epoch_end(trainer, *args, **kwargs)
+    def on_train_epoch_end(self, trainer, *args, **kwargs):
+        super().on_train_epoch_end(trainer, *args, **kwargs)
         for k, v in self.lrs.items():
             prev_value = self.last_values.get(k, None)
             new_value = v[-1]

laia/dummies/dummy_trainer.py

@@ -5,7 +5,7 @@ class DummyTrainer(pl.Trainer):
     def __init__(self, **kwargs):
         defaults = {
             "checkpoint_callback": False,
-            "logger": False,
+            "logger": True,
             "weights_summary": None,
             "max_epochs": 1,
             "limit_train_batches": 10,

requirements.txt

@@ -5,7 +5,7 @@ scipy
 matplotlib
 # cpu version: nnutils-pytorch
 nnutils-pytorch-cuda
-pytorch-lightning==1.1.0
+pytorch-lightning==1.1.7
 torch>=1.13,<1.14
 torchvision>=0.14,<0.15
 torchaudio>=0.13,<0.14

tests/callbacks/progress_bar_test.py

@@ -75,6 +75,7 @@ def test_progress_bar(tmpdir):
         r"100%\|[█]+\| 10/10 \[00:0[0-9]<00:00, "
         rf"{float_pattern}it/s, "
         rf"loss={float_pattern}, "
+        r"v_num=0, "
         rf"cer={float_pattern}%, "
         r"gpu_stats={'gpu_stats': 'baz'}]"
     )