README.md

metadata

title: clearn Demo
emoji: 🧠
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 5.23.0
python_version: '3.11'
app_file: app.py
pinned: true
license: mit
short_description: Prevent catastrophic forgetting in PyTorch

clearn Demo -- Continual Learning for PyTorch

Wrap once. Train forever.

This Space demonstrates clearn, an open-source continual learning library for PyTorch. When you fine-tune a neural network on new data, it catastrophically forgets previously learned tasks. clearn wraps any PyTorch model with strategies that prevent this forgetting.

What this demo does

Tab 1 -- Train & Inspect: Pick a continual learning strategy (EWC, SI, DER++, or GEM), configure its hyperparameters, and train a small MLP on synthetic sequential classification tasks. After training, view the model.diff() retention report and a per-task accuracy bar chart.

Tab 2 -- Compare Strategies: Run all strategies (plus a no-protection baseline) on the same data side by side. See how each strategy preserves knowledge from earlier tasks compared to naive fine-tuning.

Quick start with clearn

import clearn

model = clearn.wrap(your_model, strategy="ewc")
model.fit(task1_loader, optimizer)
model.fit(task2_loader, optimizer)
print(model.diff())

Links

• GitHub
• PyPI
• Documentation