Learning Rate Calculator
Free learning rate calculator for machine learning. Convert learning rate between decimal, scientific notation, and percentage formats. Compare Adam, SGD, and PyTorch optimizer defaults, preview linear and cosine schedules.
Loading tool...
What is Learning Rate Calculator?
The Learning Rate Calculator is a free online tool for machine learning practitioners that converts learning rates between decimal notation (0.001), scientific notation (1e-3), and percentage notation (0.1%). It provides one-click presets for common learning rates used across deep learning, loads recommended defaults for popular optimizers including Adam, AdamW, SGD, Momentum, RMSProp, and AdaGrad, and generates copy-ready PyTorch and TensorFlow/Keras code snippets with the learning rate already filled in. The tool also previews linear decay and cosine annealing schedules so you can visualize how the learning rate changes over training, and shows the Adam-family effective step-size conversion to illustrate the relationship between the nominal learning rate and the actual parameter update. All calculations run entirely in your browser with no data uploaded, making it ideal for hyperparameter tuning, model training setup, and deep learning education.
Key Benefits
- Instant conversion between decimal, scientific notation, and percentage learning rate formats
- One-click presets for common learning rates (1e-1 through 5e-5) and popular optimizer defaults
- Copy-ready PyTorch and TensorFlow/Keras code snippets with the learning rate pre-filled
- Side-by-side linear decay and cosine annealing schedule preview
- 100% client-side β works offline, no sign-up, and your input never leaves your device
Common Use Cases
- β’Hyperparameter tuning: quickly compare learning rate representations and pick the right format
- β’Model training setup: generate optimizer code for PyTorch or TensorFlow in one click
- β’Transfer learning: convert between 1e-5 and 0.00001 notation without mental math
- β’Deep learning education: explore how learning rate schedules decay over training
- β’Research workflows: understand Adam effective step-size and batch-size scaling
How to Use the Learning Rate Calculator
- Enter your learning rate: Type a learning rate in any format: decimal (0.001), scientific notation (1e-3), or percentage (0.1%). You can also click a preset button for common values.
- Pick an optimizer default (optional): Click one of the optimizer buttons (Adam, AdamW, SGD, Momentum, RMSProp, AdaGrad) to instantly load its recommended default learning rate.
- Read all format conversions: The tool shows your learning rate in decimal, scientific notation, and percentage side by side. Click the copy icon next to any value to copy it.
- Copy the code snippet: Use the PyTorch or TensorFlow/Keras snippet with the learning rate already filled in. Click copy and paste it straight into your training script.
Key Features
- Convert learning rates between decimal (0.001), scientific notation (1e-3), and percentage (0.1%) formats instantly.
- One-click presets for common learning rates (1, 1e-1, 1e-2, 1e-3, 1e-4, 5e-5) used across deep learning models.
- Quick-load optimizer defaults for Adam, AdamW, SGD, Momentum, RMSProp, and AdaGrad.
- Copy-ready PyTorch and TensorFlow/Keras code snippets with the learning rate already filled in.
- Preview linear decay and cosine annealing schedules side by side to visualize how the rate changes over training.
- Adam-family effective step-size conversion showing the impact of batch-size scaling.