smaller. faster. open.
Prune, quantize, and distill neural networks. Ship models that are smaller and faster, on any hardware.
from fasterai import optimize
result = optimize(model, sample, target='speed')
result.model # → compressed, ready to use
result.compression # → {'size': 8.1, 'latency': 3.4}
result.export('model.onnx')faster
smaller
less CO₂
Compared to the original uncompressed model. Best-case results from combined pruning, quantization, and distillation.
Choose your path
Two ways to optimize.
Use our open-source tools yourself, or let us handle it for you.
DIY
Use our tools
Done for you