max_duration=10.0,
shuffle=True,
drop_last=True,
num_buckets=2),
BucketingSampler(
CUTS,
CUTS,
max_source_duration=10.0,
shuffle=True,
drop_last=True,
num_buckets=2,
sampler_type=CutPairsSampler,
),
DynamicBucketingSampler(CUTS,
max_duration=10.0,
shuffle=True,
drop_last=True,
num_buckets=2),
DynamicBucketingSampler(CUTS,
CUTS_MOD,
max_duration=10.0,
shuffle=True,
drop_last=True,
num_buckets=2),
DynamicCutSampler(CUTS, max_duration=10.0, shuffle=True, drop_last=True),
DynamicCutSampler(CUTS,
CUTS,
max_duration=10.0,
shuffle=True,
drop_last=True),
]
assert num_sampled_cuts == expected_num_cuts
assert num_discarded_cuts == expected_discarded_cuts
SAMPLERS_FACTORIES_FOR_REPORT_TEST = [
lambda: SimpleCutSampler(DummyManifest(CutSet, begin_id=0, end_id=10)),
lambda: DynamicCutSampler(DummyManifest(CutSet, begin_id=0, end_id=10)),
lambda: CutPairsSampler(
DummyManifest(CutSet, begin_id=0, end_id=10),
DummyManifest(CutSet, begin_id=0, end_id=10),
),
lambda: BucketingSampler(DummyManifest(CutSet, begin_id=0, end_id=10),
num_buckets=2),
lambda: DynamicBucketingSampler(
DummyManifest(CutSet, begin_id=0, end_id=10),
max_duration=1.0,
num_buckets=2,
),
lambda: ZipSampler(
SimpleCutSampler(DummyManifest(CutSet, begin_id=0, end_id=10)),
SimpleCutSampler(DummyManifest(CutSet, begin_id=10, end_id=20)),
),
lambda: RoundRobinSampler(
SimpleCutSampler(DummyManifest(CutSet, begin_id=0, end_id=10)),
SimpleCutSampler(DummyManifest(CutSet, begin_id=10, end_id=20)),
),
]
@pytest.mark.parametrize("create_sampler", SAMPLERS_FACTORIES_FOR_REPORT_TEST)
def test_sampler_get_report(create_sampler):