Prevent flickering progress bar

CHANGELOG.md

@@ -269,6 +269,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed passing wrong strings for scheduler interval doesn't throw an error ([#5923](https://github.com/PyTorchLightning/pytorch-lightning/pull/5923))
 
 
+- Fixed progress bar flickering by appending 0 to floats/strings ([#6009](https://github.com/PyTorchLightning/pytorch-lightning/pull/6009))
+
+
 ## [1.1.8] - 2021-02-08
 
 ### Fixed

pytorch_lightning/callbacks/progress.py

@@ -26,12 +26,32 @@
 from typing import Optional, Union
 
 if importlib.util.find_spec('ipywidgets') is not None:
-    from tqdm.auto import tqdm
+    from tqdm.auto import tqdm as _tqdm
 else:
-    from tqdm import tqdm
+    from tqdm import tqdm as _tqdm
 
 from pytorch_lightning.callbacks import Callback
 
+_PAD_SIZE = 5
+
+
+class tqdm(_tqdm):
+    """
+    Custom tqdm progressbar where we append 0 to floating points/strings to
+    prevent the progress bar from flickering
+    """
+    @staticmethod
+    def format_num(n) -> str:
+        """ Add additional padding to the formatted numbers """
+        should_be_padded = isinstance(n, (float, str))
+        if not isinstance(n, str):
+            n = _tqdm.format_num(n)
+        if should_be_padded and 'e' not in n:
+            if '.' not in n:
+                n += '.'
+            n += "0" * (_PAD_SIZE - len(n))
+        return n
+
 
 class ProgressBarBase(Callback):
     r"""

tests/callbacks/test_progress_bar.py

@@ -20,6 +20,7 @@
 
 from pytorch_lightning import Trainer
 from pytorch_lightning.callbacks import ModelCheckpoint, ProgressBar, ProgressBarBase
+from pytorch_lightning.callbacks.progress import tqdm
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
 from tests.base import EvalModelTemplate
 from tests.helpers import BoringModel
@@ -372,3 +373,16 @@ def training_step(self, batch, batch_idx):
     pbar = trainer.progress_bar_callback.main_progress_bar
     actual = str(pbar.postfix)
     assert actual.endswith("foo=0.123, bar={'baz': tensor([1])}")
+
+
+@pytest.mark.parametrize("input_num, expected", [
+    [1, '1'],
+    [1.0, '1.000'],
+    [0.1, '0.100'],
+    [1e-3, '0.001'],
+    [1e-5, '1e-5'],
+    ['1.0', '1.000']
+])
+def test_tqdm_format_num(input_num, expected):
+    """ Check that the specialized tqdm.format_num appends 0 to floats and strings """
+    assert tqdm.format_num(input_num) == expected