[BUG] `IndexError: pop from empty list` with Rich progress bar in 14.1.0

[npurson]

When I'm using Pytorch Lightning with RichProgressBar, an error is raised:

  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/lightning/pytorch/trainer/trainer.py", line 1056, in _run_stage
    self.fit_loop.run()
  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/lightning/pytorch/loops/fit_loop.py", line 212, in run
    self.on_run_start()
  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/lightning/pytorch/loops/fit_loop.py", line 417, in on_run_start
    call._call_callback_hooks(trainer, "on_train_start")
  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/lightning/pytorch/trainer/call.py", line 227, in _call_callback_hooks
    fn(trainer, trainer.lightning_module, *args, **kwargs)
  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/lightning/pytorch/callbacks/progress/rich_progress.py", line 358, in on_train_start
    self._init_progress(trainer)
  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/lightning/pytorch/callbacks/progress/rich_progress.py", line 334, in _init_progress
    self._console.clear_live()
  File "/home/users/envs/miniconda3/envs/tmpl/lib/python3.12/site-packages/rich/console.py", line 847, in clear_live
    self._live_stack.pop()
IndexError: pop from empty list

The issue appears to be introduced in rich version 14.1.0. The code runs correctly when rich is downgraded to 14.0.0 or any lower version. The following minimal script consistently reproduces it:

import torch
from torch.utils.data import DataLoader, TensorDataset
import lightning as L
from lightning.pytorch.callbacks import RichProgressBar


class SimpleModel(L.LightningModule):
    def __init__(self):
        super().__init__()
        self.layer = torch.nn.Linear(32, 1)

    def training_step(self, batch, batch_idx):
        x, y = batch
        loss = torch.nn.functional.mse_loss(self.layer(x), y)
        self.log("train_loss", loss)
        return loss

    def configure_optimizers(self):
        return torch.optim.Adam(self.parameters(), lr=0.01)


X = torch.randn(100, 32)
y = torch.randn(100, 1)
dataset = TensorDataset(X, y)
train_loader = DataLoader(dataset, batch_size=8)

model = SimpleModel()
progress_bar = RichProgressBar()
trainer = L.Trainer(max_epochs=5, callbacks=[progress_bar], accelerator="cpu")

trainer.fit(model, train_loader)

Platform

Click to expand

> python -m rich.diagnose
╭───────────────────────── <class 'rich.console.Console'> ─────────────────────────╮
│ A high level console interface.                                                  │
│                                                                                  │
│ ╭──────────────────────────────────────────────────────────────────────────────╮ │
│ │ <console width=219 ColorSystem.TRUECOLOR>                                    │ │
│ ╰──────────────────────────────────────────────────────────────────────────────╯ │
│                                                                                  │
│     color_system = 'truecolor'                                                   │
│         encoding = 'utf-8'                                                       │
│             file = <_io.TextIOWrapper name='<stdout>' mode='w' encoding='utf-8'> │
│           height = 59                                                            │
│    is_alt_screen = False                                                         │
│ is_dumb_terminal = False                                                         │
│   is_interactive = True                                                          │
│       is_jupyter = False                                                         │
│      is_terminal = True                                                          │
│   legacy_windows = False                                                         │
│         no_color = False                                                         │
│          options = ConsoleOptions(                                               │
│                        size=ConsoleDimensions(width=219, height=59),             │
│                        legacy_windows=False,                                     │
│                        min_width=1,                                              │
│                        max_width=219,                                            │
│                        is_terminal=True,                                         │
│                        encoding='utf-8',                                         │
│                        max_height=59,                                            │
│                        justify=None,                                             │
│                        overflow=None,                                            │
│                        no_wrap=False,                                            │
│                        highlight=None,                                           │
│                        markup=None,                                              │
│                        height=None                                               │
│                    )                                                             │
│            quiet = False                                                         │
│           record = False                                                         │
│         safe_box = True                                                          │
│             size = ConsoleDimensions(width=219, height=59)                       │
│        soft_wrap = False                                                         │
│           stderr = False                                                         │
│            style = None                                                          │
│         tab_size = 8                                                             │
│            width = 219                                                           │
╰──────────────────────────────────────────────────────────────────────────────────╯
╭─── <class 'rich._windows.WindowsConsoleFeatures'> ────╮
│ Windows features available.                           │
│                                                       │
│ ╭───────────────────────────────────────────────────╮ │
│ │ WindowsConsoleFeatures(vt=False, truecolor=False) │ │
│ ╰───────────────────────────────────────────────────╯ │
│                                                       │
│ truecolor = False                                     │
│        vt = False                                     │
╰───────────────────────────────────────────────────────╯
╭────── Environment Variables ───────╮
│ {                                  │
│     'CLICOLOR': None,              │
│     'COLORTERM': 'truecolor',      │
│     'COLUMNS': None,               │
│     'JPY_PARENT_PID': None,        │
│     'JUPYTER_COLUMNS': None,       │
│     'JUPYTER_LINES': None,         │
│     'LINES': None,                 │
│     'NO_COLOR': None,              │
│     'TERM_PROGRAM': 'vscode',      │
│     'TERM': 'xterm-256color',      │
│     'TTY_COMPATIBLE': None,        │
│     'TTY_INTERACTIVE': None,       │
│     'VSCODE_VERBOSE_LOGGING': None │
│ }                                  │
╰────────────────────────────────────╯
platform="Linux"

> pip freeze | grep rich
rich==14.1.0

[github-actions]

Thank you for your issue. Give us a little time to review it.

PS. You might want to check the FAQ if you haven't done so already.

Rich was created by Will McGugan. Consider sponsoring Will's work on Rich.

This is an automated reply, generated by FAQtory

[willmcgugan]

Could you give me an MRE that doesn't rely on any external packages?

[TomJGooding]

It looks like PyTorch Lightning is calling clear_live directly in their custom ProgressBar here:

https://github.com/Lightning-AI/pytorch-lightning/blob/d25014dbda553eabe12fb0669e7451ad7aa0c576/src/lightning/pytorch/callbacks/progress/rich_progress.py#L339-L344

With the introduction of 'nested' Live instances in the latest Rich, this will cause an IndexError since this will attempt to pop from an empty 'live stack'.

[0xkarmacoma]

minimal repro:

from rich.console import Console
console = Console()
console.clear_live()

works with 14.0.0:

uv pip install rich==14.0                                                                                                                               
Resolved 4 packages in 128ms
Prepared 4 packages in 164ms
Installed 4 packages in 11ms
 + markdown-it-py==3.0.0
 + mdurl==0.1.2
 + pygments==2.19.2
 + rich==14.0.0

fails with 14.1.0:

uv pip install rich==14.1                                                        
Resolved 4 packages in 12ms
Prepared 1 package in 132ms
Installed 4 packages in 12ms
 + markdown-it-py==3.0.0
 + mdurl==0.1.2
 + pygments==2.19.2
 + rich==14.1.0

[deependujha]

Hi @npurson

I’ve pushed a fix for this in pytorch-lightning, but ideally, this PR should be merged, as it’s a cleaner and more robust solution.

Thanks,
Deependu.