view release on metacpan or  search on metacpan

t/test_module.t  view on Meta::CPAN

    # Train with original data shapes
    my $data_batch = mx->io->DataBatch(data=>[mx->nd->random_uniform(0, 9, $dshape1),
                                       mx->nd->random_uniform(5, 15, $dshape2)],
                                 label=>[mx->nd->ones($lshape)]);
    $mod->forward($data_batch);
    is_deeply($mod->get_outputs->[0]->shape, [$lshape->[0], $num_class]);
    $mod->backward();
    $mod->update();

    # Train with different batch size
    $dshape1 = [3, 3, 64, 64];
    $dshape2 = [3, 3, 32, 32];
    $lshape = [3];
    $data_batch = mx->io->DataBatch(data=>[mx->nd->random_uniform(0, 9, $dshape1),
                                       mx->nd->random_uniform(5, 15, $dshape2)],
                                 label=>[mx->nd->ones($lshape)]);
    $mod->forward($data_batch);
    is_deeply($mod->get_outputs->[0]->shape, [$lshape->[0], $num_class]);
    $mod->backward();
    $mod->update();

    $dshape1 = [20, 3, 64, 64];
    $dshape2 = [20, 3, 32, 32];
    $lshape = [20];
    $data_batch = mx->io->DataBatch(data=>[mx->nd->random_uniform(3, 5, $dshape1),
                                       mx->nd->random_uniform(10, 25, $dshape2)],
                                 label=>[mx->nd->ones($lshape)]);
    $mod->forward($data_batch);
    is_deeply($mod->get_outputs->[0]->shape, [$lshape->[0], $num_class]);
    $mod->backward();
    $mod->update();

    #Train with both different batch size and data shapes
    $dshape1 = [20, 3, 120, 120];
    $dshape2 = [20, 3, 32, 64];
    $lshape = [20];
    $data_batch = mx->io->DataBatch(data=>[mx->nd->random_uniform(0, 9, $dshape1),
                                       mx->nd->random_uniform(5, 15, $dshape2)],
                                 label=>[mx->nd->ones($lshape)]);
    $mod->forward($data_batch);
    is_deeply($mod->get_outputs->[0]->shape, [$lshape->[0], $num_class]);
    $mod->backward();
    $mod->update();

    $dshape1 = [5, 3, 28, 40];
    $dshape2 = [5, 3, 24, 16];
    $lshape = [5];
    $data_batch = mx->io->DataBatch(data=>[mx->nd->random_uniform(0, 9, $dshape1),
                                       mx->nd->random_uniform(15, 25, $dshape2)],
                                 label=>[mx->nd->ones($lshape)]);
    $mod->forward($data_batch);
    is_deeply($mod->get_outputs->[0]->shape, [$lshape->[0], $num_class]);
    $mod->backward();
    $mod->update();

    #Test score
    my $dataset_shape1 = [30, 3, 30, 30];
    my $dataset_shape2 = [30, 3, 20, 40];
    my $labelset_shape = [30];

    my $eval_dataiter = mx->io->NDArrayIter(data=>[mx->nd->random_uniform(0, 9, $dataset_shape1),
                                            mx->nd->random_uniform(15, 25, $dataset_shape2)],
                                      label=>[mx->nd->ones($labelset_shape)],
                                      batch_size=>5);
    ok(keys %{ $mod->score($eval_dataiter, 'acc') } == 1);

    #Test prediction
    $dshape1 = [1, 3, 30, 30];
    $dshape2 = [1, 3, 20, 40];
    $dataset_shape1 = [10, 3, 30, 30];
    $dataset_shape2 = [10, 3, 20, 40];

    my $pred_dataiter = mx->io->NDArrayIter(data=>[mx->nd->random_uniform(0, 9, $dataset_shape1),
                                            mx->nd->random_uniform(15, 25, $dataset_shape2)]);
    $mod->bind(data_shapes=>[['data1', $dshape1], ['data2', $dshape2]],
             for_training=>0, force_rebind=>1);
    is_deeply($mod->predict($pred_dataiter)->shape, [10, $num_class]);

}

test_module_input_grads();
test_module_dtype();
test_monitor();
test_module_switch_bucket();
test_module_layout();
test_module_states();
test_module_reshape();
test_save_load();
test_executor_group();
test_module_set_params();
test_forward_reshape();