Personal project, not a company
Reference implementation running
EnAi
Cut AI training compute at runtime.
A training run keeps computing gradients for layers that stopped learning hours ago. EnAi watches gradient energy per layer while the run is in flight, freezes each layer the moment its contribution collapses, and raises the batch size into the headroom that frees up. All of it happens mid-run, with no restart and no change to the model code.
Solo. I build the engine, the study and the measurement.
- 6.91%
- training compute removedAnalytic ledger, confirmed by autograd
- +20.9%
- throughput once engagedMachine normalised, median of 12 runs
- 1,563 to 0
- backward passes into frozen layersCounted per layer by hooks left attached
- 92%
- of the theoretical ceiling reached+7.42% predicted, +6.85% measured
- PyTorch
- Python
- Autograd hooks
- Metal Performance Shaders
- ResNet-18 / CIFAR-10
enai-lab.comRead the detail