def test_to_backend_with_tf_and_pytorch():
try:
import torch
import tensorflow as tf
except ImportError:
print("Pytorch hasn't been imported and tested")
return
tf.compat.v1.enable_eager_execution()
ds = dataset.load("mnist/mnist")
tfds = ds.to_tensorflow()
ptds = ds.to_pytorch()
ptds = torch.utils.data.DataLoader(
ptds,
batch_size=1,
num_workers=1,
collate_fn=ds.collate_fn if "collate_fn" in dir(ds) else None,
)
for i, (batchtf, batchpt) in enumerate(zip(tfds, ptds)):
gt = ds["labels"][i].compute()
assert gt == batchtf["labels"].numpy()
assert gt == batchpt["labels"].numpy()
if i > 10:
break
def test_to_backend_with_tf_and_pytorch_multiworker():
import tensorflow as tf
import torch
tf.compat.v1.enable_eager_execution()
ds = dataset.load("mnist/mnist")
tfds = ds.to_tensorflow().batch(8)
ptds = ds.to_pytorch()
ptds = torch.utils.data.DataLoader(
ptds,
batch_size=8,
num_workers=8,
collate_fn=ds.collate_fn if "collate_fn" in dir(ds) else None,
)
for i, (batchtf, batchpt) in enumerate(zip(tfds, ptds)):
assert np.all(batchtf["labels"].numpy() == batchpt["labels"].numpy())
if i > 10:
break
def test_tensor_dtag():
t = tensor.from_array(np.array([1, 2], dtype="int32"), dtag="image")
ds = dataset.from_tensors({"name": t})
ds.store("./data/new/test")
ds = dataset.load("./data/new/test")
assert ds["name"].dtag == "image"
