ContinualAI/avalanche

Health v2 · maintenance only

Adoption not part of the score

2088 stars · 319 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Avalanche is an end-to-end continual learning library built on PyTorch, developed by ContinualAI. It provides modules for benchmarks, training strategies, and evaluation metrics for reproducible continual/lifelong learning research.

Use cases

• run continual learning experiments in pytorch
• benchmark continual learning algorithms
• reproduce continual learning paper results
• train models incrementally without catastrophic forgetting
• evaluate lifelong learning strategies with standard metrics
• create custom task streams from my own datasets

When to choose

• you do continual or lifelong learning research on PyTorch
• you need standardized CL benchmarks, strategies, and metrics
• you want reproducible baselines for CL papers

When to avoid

• you need standard supervised training without incremental data streams
• you work outside PyTorch (e.g., TensorFlow/JAX)
• you need production ML serving rather than research prototyping

Facets

library · maturity active

machine-learning deep-learning benchmarking llm-training machine-learning deep-learning python cross-platform continual-learning lifelong-learning pytorch catastrophic-forgetting benchmarks research gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ContinualAI/avalanche")