[source]

Dataset

pyjet.data.Dataset()

An abstract container for data designed to be passed to a model. This container should implement create_batch. It is only necessary to implement validation_split() if you use this module to split your data into a train and test set. Same goes for kfold()

• _Note_:

Though not forced, a Dataset is really a constant object. Once created, it should not be mutated in any way.

---

[source]

NpDataset

pyjet.data.NpDataset(x, y=None)

A Dataset that is built from numpy data.

Arguments

x -- The input data as a numpy array y -- The target data as a numpy array (optional)

---

[source]

HDF5Dataset

pyjet.data.HDF5Dataset(x, y=None)

---

[source]

TorchDataset

pyjet.data.TorchDataset(x, y=None)

---

[source]

DatasetGenerator

pyjet.data.DatasetGenerator(dataset, steps_per_epoch=None, batch_size=None, shuffle=True, seed=None)

An iterator to create batches for a model using a Dataset. 2 of the following must be defined -- The input Dataset's length -- steps_per_epoch -- batch_size Also, if the Dataset's length is not defined, its create_batch method should not take any inputs

Arguments

dataset -- the dataset to generate from steps_per_epoch -- The number of iterations in one epoch (optional) batch_size -- The number of samples in one batch (optional) shuffle -- Whether or not to shuffle the dataset before each epoch - default: True seed -- A seed for the random number generator (optional).