ai-agent

pytorch-lr-finder: Optimize Learning Rates

Name: pytorch-lr-finder: Optimize Learning Rates
Author: davidtvs

Find optimal learning rates with pytorch-lr-finder, a tool for founders using PyTorch, backed by 1.0k+ GitHub stars

From the repo · “A learning rate range test implementation in PyTorch”

1,009 stars122 forksPythonGuide quality 8/10 · reviewed Aug 6Updated 6/24/2025100% free · open source

What it does

Automates the process of finding the optimal learning rate for a PyTorch model by performing a learning rate range test

When to use it

•When training a new PyTorch model and unsure about the ideal learning rate
•When experiencing underfitting or overfitting issues due to a suboptimal learning rate
•When comparing the performance of different PyTorch models with varying learning rates

Ready-to-paste prompt

lr_finder = LRFinder(model=resnet18(), optimizer=torch.optim.SGD(model.parameters(), lr=1e-6)); lr_finder.find(learning_rate=1e-6, end_lr=1, num_iter=100); lr_finder.plot()

Heads up: Make sure to specify the correct model and optimizer when initializing the LRFinder instance, as the learning rate range test will be performed on the provided model and optimizer

Saves to your device

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

⚡ Runs out of the box

A prompt/skill package — nothing to install beyond adding it to your AI tool.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/pytorch-lr-finder && curl -fsSL https://workflowstacks.com/api/skills/pytorch-lr-finder/claude-skill -o ~/.claude/skills/pytorch-lr-finder/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

How pytorch-lr-finder: Optimize Learning Rates works

Codeflow

Free to inspect

Pytorch-lr-finder: Optimize Learning Rates is mostly documents (5 doc files, 5 small scripts) — something you read, not something you run. There is nothing to install. Last commit 16 months ago, MIT license.

Size

Mostly documents

5 documents · 5 small scripts · 2 min skim

Setup

Nothing to install

A course / notebook collection. Read it; run cells if you like.

Runs on

Nowhere — you read it

A reading resource, not a program.

Python 100%

Where to start reading

1
README.md
Start here — what it does and how to install it
2
examples/README.md
A worked example — copy this to get going

What's in each folder

examples/Examples you can copy8 files

torch_lr_finder/Folder2 files

.github/CI / automation (GitHub Actions)7 files

tests/Tests — proof it works6 files

images/Images & static assets1 files

READMEHas testsDocumentedCI checksExamples includedMIT licenseLast update 16 mo ago

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

⚡ Runs out of the box

A prompt/skill package — nothing to install beyond adding it to your AI tool.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/pytorch-lr-finder && curl -fsSL https://workflowstacks.com/api/skills/pytorch-lr-finder/claude-skill -o ~/.claude/skills/pytorch-lr-finder/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

Quick Actions

Details

Creator

davidtvs

Language

Python