def test_append_dataset():
dt = {"first": Tensor(shape=(250, 300)), "second": "float"}
url = "./data/test/model"
ds = Dataset(schema=dt, shape=(100,), url=url, mode="w")
ds.append_shape(20)
ds["first"][0] = np.ones((250, 300))
assert len(ds) == 120
assert ds["first"].shape[0] == 120
assert ds["first", 5:10].shape[0] == 5
assert ds["second"].shape[0] == 120
ds.commit()
ds = Dataset(url)
assert ds["first"].shape[0] == 120
assert ds["first", 5:10].shape[0] == 5
assert ds["second"].shape[0] == 120
def upload(self, results, url: str, token: dict, progressbar: bool = True):
"""Batchified upload of results
For each tensor batchify based on its chunk and upload
If tensor is dynamic then still upload element by element
Parameters
----------
dataset: hub.Dataset
Dataset object that should be written to
results:
Output of transform function
progressbar: bool
Returns
----------
ds: hub.Dataset
Uploaded dataset
"""
shape = (len(list(results.values())[0]), )
ds = Dataset(
url,
mode="w",
shape=shape, # unkownn
schema=self.schema,
token=token,
cache=False,
)
tasks = []
for key, value in results.items():
length = ds[key].chunksize[0]
batched_values = batchify(value, length)
chunk_id = list(range(len(batched_values)))
index_batched_values = list(zip(chunk_id, batched_values))
results = [
self.upload_chunk.remote(el, key=key, ds=ds)
for el in index_batched_values
]
tasks.extend(results)
ray.get(tasks)
ds.commit()
return ds
def main():
# Tag is set {Username}/{Dataset}
tag = "davitb/basic11"
# Create dataset
ds = Dataset(
tag,
shape=(4, ),
schema={
"image": schema.Tensor((512, 512), dtype="float"),
"label": schema.Tensor((512, 512), dtype="float"),
},
)
# Upload Data
ds["image"][:] = np.ones((4, 512, 512))
ds["label"][:] = np.ones((4, 512, 512))
ds.commit()
# Load the data
ds = Dataset(tag)
print(ds["image"][0].compute())
def main():
schema = {
"image": Image(shape=(None, None), max_shape=(28, 28)),
"label": ClassLabel(num_classes=10),
}
path = "./data/examples/new_api_intro2"
ds = Dataset(path, shape=(10, ), mode="w", schema=schema)
print(len(ds))
for i in range(len(ds)):
with Timer("writing single element"):
ds["image", i] = np.ones((28, 28), dtype="uint8")
ds["label", i] = 3
ds.resize_shape(200)
print(ds.shape)
print(ds["label", 100:110].numpy())
with Timer("Committing"):
ds.commit()
ds = Dataset(path)
print(ds.schema)
print(ds["image", 0].compute())
def upload(
self,
results,
url: str,
token: dict,
progressbar: bool = True,
public: bool = True,
):
"""Batchified upload of results.
For each tensor batchify based on its chunk and upload.
If tensor is dynamic then still upload element by element.
Parameters
----------
dataset: hub.Dataset
Dataset object that should be written to
results:
Output of transform function
progressbar: bool
public: bool, optional
only applicable if using hub storage, ignored otherwise
setting this to False allows only the user who created it to access the dataset and
the dataset won't be visible in the visualizer to the public
Returns
----------
ds: hub.Dataset
Uploaded dataset
"""
if len(list(results.values())) == 0:
shape = (0, )
else:
shape = (len(list(results.values())[0]), )
ds = Dataset(
url,
mode="w",
shape=shape,
schema=self.schema,
token=token,
cache=False,
public=public,
)
tasks = []
for key, value in results.items():
length = ds[key].chunksize[0]
value = get_value(value)
value = str_to_int(value, ds.tokenizer)
batched_values = batchify(value, length)
chunk_id = list(range(len(batched_values)))
index_batched_values = list(zip(chunk_id, batched_values))
ds._tensors[f"/{key}"].disable_dynamicness()
results = [
self.upload_chunk.remote(el, key=key, ds=ds)
for el in index_batched_values
]
tasks.extend(results)
results = ray.get(tasks)
self.set_dynamic_shapes(results, ds)
ds.commit()
return ds
