Benchmarks
Benchmarks and DatasetCode Examples
Last updated
Was this helpful?
Benchmarks and DatasetCode Examples
Last updated
Was this helpful?
Avalanche offers significant support for defining your own benchmarks (instantiation of one scenario with one or multiple datasets) or using "classic" benchmarks already consolidate in the literature.
You can find examples related to the benchmarks here:
: in this simple example we show all the different ways you can use MNIST with Avalanche.
: in this example a CIFAR100 is used with its canonical split in 10 experiences, 10 classes each.
: training and evaluating on CLEAR benchmark (RGB images)
: Training and evaluating on CLEAR benchmark (with pre-trained features)
: about the utils you can use create a detection benchmark.
: this example makes use of the Endless-Continual-Learning-Simulator's derived dataset scenario.
: In this example we show a simple way to use the ctrl benchmark.
: this example trains on Split CIFAR10 with Naive strategy. In this example each experience has a different task label.