Show current learning rate in training progress bar by Copilot · Pull Request #809 · roboflow/rf-detr

Copilot · 2026-03-17T11:22:29Z

Users monitoring LR during fine-tuning had no visibility into the current learning rate from the tqdm progress bar — making it hard to observe warmup ramp-up, step decay, or cosine annealing in real time.

Changes

RFDETRModule.training_step: Reads the current LR from optimizer.param_groups[0]["lr"] after each step and logs it as train/lr with prog_bar=True, on_step=True, on_epoch=False. Works correctly for both step-decay and cosine schedulers.
test_module.py: Adds test_logs_learning_rate_to_prog_bar to TestTrainingStep asserting train/lr is logged with prog_bar=True.
test_metrics_csv.py: Adds train/lr to _REQUIRED_DETECTION to ensure it's always present in metrics.csv.

The LR now appears alongside train/loss in the progress bar:

Epoch 1: 100%|████| loss=2.34 train/lr=1e-04

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

pypi.nvidia.com
- Triggering command: `/home/REDACTED/.cache/uv/builds-v0/.tmp4iUnUY/bin/python /home/REDACTED/.cache/uv/builds-v0/.tmp4iUnUY/bin/python -c import sys

if sys.path[0] == "":
sys.path.pop(0)

sys.path = [] + sys.path

import wheel_stub.buildapi as backend

wheel_filename = backend.build_wheel("/home/REDACTED/.cache/uv/builds-v0/.tmpt5WLug", {}, None)
with open("/home/REDACTED/.cache/uv/builds` (dns block)

Triggering command: `/home/REDACTED/.cache/uv/builds-v0/.tmppOU1HV/bin/python /home/REDACTED/.cache/uv/builds-v0/.tmppOU1HV/bin/python -c import sys

if sys.path[0] == "":
sys.path.pop(0)

sys.path = [] + sys.path

import wheel_stub.buildapi as backend

wheel_filename = backend.build_wheel("/home/REDACTED/.cache/uv/builds-v0/.tmpmKS7yB", {}, None)
with open("/home/REDACTED/.cache/uv/builds` (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

Configure Actions setup steps to set up my environment, which run before the firewall is enabled

Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

Original prompt

This section details on the original issue you should resolve

<issue_title>Version 1.6.0rc01: we need to see current learning_rate in progess bar</issue_title>
<issue_description>### Search before asking

I have searched the RF-DETR issues and found no similar feature requests.

Description

I tried the new 1.6.0rc01. It look neat but i need to monitor current learning_rate that will help me finetune the lr value when i use pretrained_weights. Please consider

Use case

No response

Additional

No response

Are you willing to submit a PR?

Yes I'd like to help by submitting a PR!</issue_description>

Comments on the Issue (you are @copilot in this section)

Fixes Version 1.6.0rc: we need to see current learning_rate in progess bar #804

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.

codecov · 2026-03-17T11:29:21Z

Codecov Report

❌ Patch coverage is 90.90909% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 75%. Comparing base (0755e60) to head (4e0d532).
⚠️ Report is 1 commits behind head on develop.

Additional details and impacted files

@@          Coverage Diff           @@
##           develop   #809   +/-   ##
======================================
  Coverage       75%    75%           
======================================
  Files           91     91           
  Lines         6806   6817   +11     
======================================
+ Hits          5079   5089   +10     
- Misses        1727   1728    +1

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Co-authored-by: Borda <6035284+Borda@users.noreply.github.com>

Copilot

Pull request overview

Adds live learning-rate visibility during fine-tuning by logging the optimizer LR as a Lightning metric so it can appear in the tqdm progress bar.

Changes:

Log train/lr from the optimizer’s param group during RFDETRModule.training_step with prog_bar=True.
Add a unit test asserting train/lr is logged with prog_bar=True.
Require train/lr to exist in metrics.csv via the metrics CSV integration test.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
`src/rfdetr/training/module.py`	Logs `train/lr` each training step for progress-bar visibility.
`tests/training/test_module.py`	Adds a unit test validating `train/lr` is logged to the progress bar.
`tests/training/test_metrics_csv.py`	Adds `train/lr` to the required `metrics.csv` columns set.

Comments suppressed due to low confidence (1)

tests/training/test_metrics_csv.py:83

test_metrics_csv.py is described as enforcing the metric columns that plot_metrics() needs, but plot_metrics() currently only reads train/loss, val/loss, and test/loss (plus val/* AP/AR metrics). Adding train/lr here makes the test intent/documentation inconsistent and creates extra coupling; either update plot_metrics() to actually consume/plot LR, or keep this required-set limited to what the plotting code uses.

_REQUIRED_DETECTION = frozenset(
    {
        "train/loss",
        "train/lr",
        "val/loss",
        "val/mAP_50",
        "val/mAP_50_95",
        "val/mAR",
    }

You can also share your feedback on Copilot code review. Take the survey.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

- Updated `train/lr` logging to always set `on_step=True` and `on_epoch=False` in training module. - Added assertions in tests to validate these settings.

Copilot

Pull request overview

This PR adds visibility into the current learning rate during training by logging it to the Lightning progress bar (and ensuring it appears in metrics.csv), which helps users monitor scheduler behavior (warmup/decay/cosine) while fine-tuning.

Changes:

Log the optimizer learning rate as train/lr from RFDETRModule.training_step with prog_bar=True and step-level logging.
Add a unit test asserting train/lr is logged with progress-bar visibility.
Extend the metrics.csv integration test required-column set to include train/lr.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
`src/rfdetr/training/module.py`	Logs current LR as `train/lr` during `training_step` for progress-bar visibility.
`tests/training/test_module.py`	Adds a training-step unit test asserting `train/lr` is logged with `prog_bar=True` and step-only flags.
`tests/training/test_metrics_csv.py`	Requires `train/lr` to exist and have non-NaN values in `metrics.csv`.

Comments suppressed due to low confidence (1)

tests/training/test_metrics_csv.py:83

_REQUIRED_DETECTION is described as the set of columns that plot_metrics() needs, but plot_metrics() currently only plots loss + val/* AP/AR metrics and does not use train/lr. Either update the test/module comments to reflect that this is an additional required metric beyond what plot_metrics() consumes, or extend plot_metrics() to handle/plot train/lr so the test intent matches the visualization behavior.

_REQUIRED_DETECTION = frozenset(
    {
        "train/loss",
        "train/lr",
        "val/loss",
        "val/mAP_50",
        "val/mAP_50_95",
        "val/mAR",
    }

You can also share your feedback on Copilot code review. Take the survey.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

- Added a real optimizer to tests for `param_groups` to ensure realistic "lr" key handling. - Mocked optimizer return to align with the updated test setup.

Initial plan

62af315

Copilot AI assigned Copilot and Borda Mar 17, 2026

Copilot started work on behalf of Borda March 17, 2026 11:22 View session

feat: show current learning rate in training progress bar

8edf57b

Co-authored-by: Borda <6035284+Borda@users.noreply.github.com>

Copilot AI changed the title ~~[WIP] [ISSUE-1] Add current learning rate to progress bar~~ Show current learning rate in training progress bar Mar 17, 2026

Copilot AI requested a review from Borda March 17, 2026 11:33

Copilot finished work on behalf of Borda March 17, 2026 11:33

Borda marked this pull request as ready for review March 17, 2026 11:38

Borda requested review from SkalskiP, isaacrob and probicheaux as code owners March 17, 2026 11:38

Copilot AI review requested due to automatic review settings March 17, 2026 11:38

Copilot started reviewing on behalf of Borda March 17, 2026 11:38 View session

Borda approved these changes Mar 17, 2026

View reviewed changes

Copilot AI reviewed Mar 17, 2026

View reviewed changes

Comment thread src/rfdetr/training/module.py Outdated

Comment thread tests/training/test_module.py

Borda and others added 2 commits March 17, 2026 12:44

Potential fix for pull request finding

3b94c3e

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

test: ensure on_step and on_epoch are explicitly set for LR logging

9eb74a7

- Updated `train/lr` logging to always set `on_step=True` and `on_epoch=False` in training module. - Added assertions in tests to validate these settings.

Borda requested a review from Copilot March 17, 2026 12:05

Copilot started reviewing on behalf of Borda March 17, 2026 12:06 View session

Copilot AI reviewed Mar 17, 2026

View reviewed changes

Comment thread src/rfdetr/training/module.py Outdated

Comment thread src/rfdetr/training/module.py Outdated

Borda and others added 2 commits March 17, 2026 13:15

Apply suggestions from code review

87a77aa

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

test(module): ensure real optimizer is used in tests

4e0d532

- Added a real optimizer to tests for `param_groups` to ensure realistic "lr" key handling. - Mocked optimizer return to align with the updated test setup.

Borda merged commit 4087a88 into develop Mar 17, 2026
23 checks passed

Borda deleted the copilot/add-learning-rate-progress-bar branch March 17, 2026 14:04

Navigation Menu

Show current learning rate in training progress bar - #809

Show current learning rate in training progress bar#809
Borda merged 6 commits into
developfrom
copilot/add-learning-rate-progress-bar

Copilot AI commented Mar 17, 2026 •

edited

Loading

Uh oh!

codecov Bot commented Mar 17, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

Copilot AI commented Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes

I tried to connect to the following addresses, but was blocked by firewall rules:

Description

Use case

Additional

Are you willing to submit a PR?

Comments on the Issue (you are @copilot in this section)

Uh oh!

codecov Bot commented Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Copilot AI commented Mar 17, 2026 •

edited

Loading

codecov Bot commented Mar 17, 2026 •

edited

Loading