cartpole_gym_replay.py

import pickle

import datasets
import numpy as np

# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_BASE_URL = "https://huggingface.co/datasets/marktrovinger/cartpole_gym_replay/resolve/main"
_DATA_URL = f"{_BASE_URL}/replay_buffer_npz.npz"

_DESCRIPTION = """ \
Testing a cartpole replay.
"""

_HOMEPAGE = "blah"

_LICENSE = "MIT"


class DecisionTransformerGymDataset(datasets.GeneratorBasedBuilder):
    def _info(self):

        features = datasets.Features(
            {
                "observations": datasets.Sequence(datasets.Sequence(datasets.Value("float32"))),
                "actions": datasets.Sequence(datasets.Sequence(datasets.Value("float32"))),
                "rewards": datasets.Sequence(datasets.Value("float32")),
                "dones": datasets.Sequence(datasets.Value("float32")),
                # These are the features of your dataset like images, labels ...
            }
        )

        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # This defines the different columns of the dataset and their types
            # Here we define them above because they are different between the two configurations
            features=features,
            # If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
            # specify them. They'll be used if as_supervised=True in builder.as_dataset.
            # supervised_keys=("sentence", "label"),
            # Homepage of the dataset for documentation
            homepage=_HOMEPAGE,
            # License for the dataset if available
            license=_LICENSE,
        )
    def _split_generators(self, dl_manager):
        urls = _DATA_URL
        data_dir = dl_manager.download_and_extract(urls)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "filepath": data_dir,
                    "split": "train",
                },
            )
        ]
    # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
    def _generate_examples(self, filepath, split):
        
        #trajectories = pickle.load(f)
        #traj = np.load(filepath)
        obs = np.load(f'{filepath}/observations.npy')
        act = np.load(f'{filepath}/actions.npy')
        rew = np.load(f'{filepath}/rewards.npy')
        dones = np.load(f'{filepath}/dones.npy')
        for idx, value in enumerate(obs):
            yield idx, {
                "observations": obs[i],
                "actions": act[i],
                "rewards": rew[i],
                "dones": done[i],
            }