{ "featuresDict": { "features": { "episode_metadata": { "featuresDict": { "features": { "file_path": { "description": "Path to the original data file.", "pythonClassName": "tensorflow_datasets.core.features.text_feature.Text", "text": {} }, "has_language": { "description": "True if language exists in observation, otherwise empty string.", "pythonClassName": "tensorflow_datasets.core.features.scalar.Scalar", "tensor": { "dtype": "bool", "encoding": "none", "shape": {} } }, "object_list": { "description": "List of object in the scene.", "pythonClassName": "tensorflow_datasets.core.features.text_feature.Text", "text": {} }, "robot_id": { "description": "Robot ID.", "pythonClassName": "tensorflow_datasets.core.features.text_feature.Text", "text": {} }, "success": { "description": "True if the episode is deemed successful by VLM, otherwise False.", "pythonClassName": "tensorflow_datasets.core.features.scalar.Scalar", "tensor": { "dtype": "bool", "encoding": "none", "shape": {} } }, "task_list": { "description": "List of language task that is available for the scene.", "pythonClassName": "tensorflow_datasets.core.features.text_feature.Text", "text": {} }, "time": { "description": "Time the episode is recorded.", "pythonClassName": "tensorflow_datasets.core.features.text_feature.Text", "text": {} } } }, "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict" }, "steps": { "pythonClassName": "tensorflow_datasets.core.features.dataset_feature.Dataset", "sequence": { "feature": { "featuresDict": { "features": { "action": { "description": "Robot action, consists of [3x XYZ delta, 3x roll-pitch-yaw delta, 1x gripper absolute].", "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", "tensor": { "dtype": "float32", "encoding": "none", "shape": { "dimensions": [ "7" ] } } }, "goal": { "description": "Goal image. Generated by SuSIE.", "image": { "dtype": "uint8", "encodingFormat": "jpeg", "shape": { "dimensions": [ "256", "256", "3" ] } }, "pythonClassName": "tensorflow_datasets.core.features.image_feature.Image" }, "is_first": { "description": "True on first step of the episode.", "pythonClassName": "tensorflow_datasets.core.features.scalar.Scalar", "tensor": { "dtype": "bool", "encoding": "none", "shape": {} } }, "is_last": { "description": "True on last step of the episode.", "pythonClassName": "tensorflow_datasets.core.features.scalar.Scalar", "tensor": { "dtype": "bool", "encoding": "none", "shape": {} } }, "language_instruction": { "description": "Language Instruction.", "pythonClassName": "tensorflow_datasets.core.features.text_feature.Text", "text": {} }, "observation": { "featuresDict": { "features": { "image_0": { "description": "Main camera RGB observation (fixed position).", "image": { "dtype": "uint8", "encodingFormat": "jpeg", "shape": { "dimensions": [ "256", "256", "3" ] } }, "pythonClassName": "tensorflow_datasets.core.features.image_feature.Image" }, "state": { "description": "Robot end effector state, consists of [3x XYZ, 3x roll-pitch-yaw, 1x gripper]", "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", "tensor": { "dtype": "float32", "encoding": "none", "shape": { "dimensions": [ "7" ] } } } } }, "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict" } } }, "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict" }, "length": "-1" } } } }, "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict" }