Comparison between Classical and LeafLib model — evaluating training performance, efficiency, and optimization potential.
Published: October 2025
Dataset Volume
30,000 × 41 samples
Training GPU
Apple M1 Pro (16-core GPU, Metal / MPS acceleration)
Model Parameters
Classical Model: ~ 9.4k
LeafNet Model: ~ 1.4k
Accuracy
Classical Model: 83% (After 7th epoch)
LeafNet Model: 100% (After 2st epoch)
Training Time
Classical Model: ~0.2s per train
LeafNet Model: ~4s per train
1. Training time for LeafNet v1.0 is not fully optimized and will be improved in future versions. The next step is to make time consuming computation in low-level programing language instead of Python.
2. LeafNet has logarithmic parameters grouth that is why it will always smaller amout of parameters.