black version 23.1 format updates (facebookresearch#1110)

habitat-baselines/habitat_baselines/common/construct_vector_env.py

@@ -75,7 +75,6 @@ def construct_envs(
     for i in range(num_environments):
         proc_config = config.copy()
         with read_write(proc_config):
-
             task_config = proc_config.habitat
             task_config.seed = task_config.seed + i
             if len(scenes) > 0:

habitat-baselines/habitat_baselines/common/obs_transformers.py

@@ -785,7 +785,6 @@ def calculate_zfactor(
     def forward(
         self, batch: torch.Tensor, is_depth: bool = False
     ) -> torch.Tensor:
-
         # Depth conversion for input tensors
         if is_depth and self.input_zfactor is not None:
             input_b = batch.size()[0] // self.input_len
@@ -931,7 +930,6 @@ def transform_observation_space(
     def forward(
         self, observations: Dict[str, torch.Tensor]
     ) -> Dict[str, torch.Tensor]:
-
         for i, target_sensor_uuid in enumerate(self.target_uuids):
             # number of input and input sensor uuids
             in_len = self.converter.input_len
@@ -1185,7 +1183,6 @@ def __init__(
 
     @classmethod
     def from_config(cls, config):
-
         if hasattr(config, "target_uuids"):
             # Optional Config Value to specify target uuid
             target_uuids = config.target_uuids

habitat-baselines/habitat_baselines/common/rollout_storage.py

@@ -72,7 +72,6 @@ def __init__(
             numsteps + 1, num_envs, *action_shape
         )
         if discrete_actions:
-
             assert isinstance(self.buffers["actions"], torch.Tensor)
             assert isinstance(self.buffers["prev_actions"], torch.Tensor)
             self.buffers["actions"] = self.buffers["actions"].long()

habitat-baselines/habitat_baselines/il/data/data.py

@@ -111,7 +111,6 @@ def __init__(
                         self.scene_episode_dict[scene],
                         desc="Saving episode frames for each scene",
                     ):
-
                         if self.only_vqa_task:
                             pos_queue = episode.shortest_paths[0][
                                 -self.num_frames :  # noqa: E203

habitat-baselines/habitat_baselines/il/data/eqa_cnn_pretrain_data.py

@@ -98,7 +98,6 @@ def save_frames(self, pos_queue: List[ShortestPathPoint]) -> None:
         """
 
         for pos in pos_queue:
-
             observation = self.env.sim.get_observations_at(
                 pos.position, pos.rotation
             )

habitat-baselines/habitat_baselines/il/data/nav_data.py

@@ -280,7 +280,6 @@ def preprocess_actions(self) -> None:
         0 - NULL; 1 - START; 2 - FWD; 3 - LEFT; 4 - RIGHT; 5 - STOP;
         """
         for ep in self.episodes:
-
             ep.actions = [x.action + 2 for x in ep.shortest_paths[0]]
             ep.action_length = len(ep.actions)
             (

habitat-baselines/habitat_baselines/il/metrics.py

@@ -64,7 +64,6 @@ def update(self, values: List) -> None:
         self.stats.append(copy.deepcopy(current_stats))
 
     def get_stat_string(self, mode: int = 1) -> str:
-
         stat_string = ""
 
         for k, v in self.info.items():
@@ -88,7 +87,6 @@ def get_stats(self, mode: int = 1) -> List[float]:
         return stats
 
     def dump_log(self) -> bool:
-
         if self.log_json is None:
             return False
 

habitat-baselines/habitat_baselines/il/models/models.py

@@ -325,7 +325,6 @@ def __init__(
     def forward(
         self, images: Tensor, questions: Tensor
     ) -> Tuple[Tensor, Tensor]:
-
         N, T, _, _, _ = images.size()
         # bs x 5 x 3 x 256 x 256
         img_feats = self.cnn(
@@ -446,7 +445,6 @@ def forward(
         controller_actions_in: Tensor,
         controller_action_lengths: Tensor,
     ) -> Tuple[Tensor, Tensor, Tensor]:
-
         N_p, T_p, _ = planner_img_feats.size()
 
         planner_img_feats = self.cnn_fc_layer(planner_img_feats)
@@ -528,7 +526,6 @@ def planner_step(
     def controller_step(
         self, img_feats: Tensor, actions_in: Tensor, hidden_in: Tensor
     ) -> Tensor:
-
         img_feats = self.cnn_fc_layer(img_feats)
         actions_embed = self.planner_nav_rnn.action_embed(actions_in)
 
@@ -647,7 +644,6 @@ def forward(
         action_lengths: Tensor,
         hidden: bool = False,
     ) -> Union[Tuple[Tensor, Tensor], Tuple[Tensor, Tensor, Tensor]]:
-
         T: Union[int, bool] = False
         if self.image_input is True:
             N, T, _ = img_feats.size()
@@ -695,7 +691,6 @@ def step_forward(
         actions_in: Tensor,
         hidden: HiddenState,
     ) -> Tuple[Tensor, Tensor]:
-
         T: Union[bool, int] = False
         if self.image_input is True:
             N, T, _ = img_feats.size()

habitat-baselines/habitat_baselines/il/trainers/eqa_cnn_pretrain_trainer.py

@@ -268,7 +268,6 @@ def _eval_checkpoint(
                     % config.habitat_baselines.il.eval_save_results_interval
                     == 0
                 ):
-
                     result_id = "ckpt_{}_{}".format(
                         checkpoint_index, idx[0].item()
                     )

habitat-baselines/habitat_baselines/il/trainers/pacman_trainer.py

@@ -64,7 +64,6 @@ def _save_nav_results(
         writer: TensorboardWriter,
         video_option: list,
     ) -> None:
-
         r"""For saving NAV-PACMAN eval results.
         Args:
             ckpt_path: path of checkpoint being evaluated
@@ -444,7 +443,6 @@ def _eval_checkpoint(
                             )
 
                             for step in range(planner_actions_in.size(0)):
-
                                 (
                                     planner_scores,
                                     planner_hidden,

habitat-baselines/habitat_baselines/il/trainers/vqa_trainer.py

@@ -62,7 +62,6 @@ def _save_vqa_results(
         q_vocab_dict: VocabDict,
         ans_vocab_dict: VocabDict,
     ) -> None:
-
         r"""For saving VQA results.
         Args:
             ckpt_idx: idx of checkpoint being evaluated
@@ -395,7 +394,6 @@ def _eval_checkpoint(
                     % config.habitat_baselines.il.eval_save_results_interval
                     == 0
                 ):
-
                     self._save_vqa_results(
                         checkpoint_index,
                         episode_ids,

habitat-baselines/habitat_baselines/rl/ddppo/algo/ddppo.py

@@ -98,6 +98,7 @@ def init_distributed(self, find_unused_params: bool = True) -> None:
                                    forward pass, otherwise the gradient reduction
                                    will not work correctly.
         """
+
         # NB: Used to hide the hooks from the nn.Module,
         # so they don't show up in the state_dict
         class Guard:  # noqa: SIM119

habitat-baselines/habitat_baselines/rl/hrl/hierarchical_policy.py

@@ -172,7 +172,6 @@ def act(
         masks,
         deterministic=False,
     ):
-
         self._high_level_policy.apply_mask(masks)  # type: ignore[attr-defined]
 
         should_terminate: torch.BoolTensor = torch.zeros(

habitat-baselines/habitat_baselines/rl/hrl/skills/art_obj.py

@@ -40,7 +40,6 @@ def on_enter(
     def _is_skill_done(
         self, observations, rnn_hidden_states, prev_actions, masks, batch_idx
     ) -> torch.BoolTensor:
-
         cur_resting_pos = observations[RelativeRestingPositionSensor.cls_uuid]
 
         did_leave_start_zone = (

habitat-baselines/habitat_baselines/rl/models/action_embedding.py

@@ -24,6 +24,7 @@ class BoxActionEmbedding(nn.Module):
     x = (a - low) / (high - low). This assumes that the bounds
     in the action space are tight
     """
+
     # NeRF style sinusoidal embedding for continuous actions
     def __init__(self, action_space: gym.spaces.Box, dim_per_action: int = 32):
         super().__init__()
@@ -113,7 +114,6 @@ def __init__(self, action_space: ActionSpace, dim_per_action: int = 32):
             )
             self.embedding_slices.append(slice(0, 1))
         else:
-
             ptr = 0
             for space in iterate_action_space_recursively(action_space):
                 if isinstance(space, gym.spaces.Box):

habitat-baselines/habitat_baselines/rl/ppo/ppo.py

@@ -62,7 +62,6 @@ def __init__(
         entropy_target_factor: float = 0.0,
         use_adaptive_entropy_pen: bool = False,
     ) -> None:
-
         super().__init__()
 
         self.actor_critic = actor_critic
@@ -159,7 +158,6 @@ def update(
         self,
         rollouts: RolloutStorage,
     ) -> Dict[str, float]:
-
         advantages = self.get_advantages(rollouts)
 
         learner_metrics = collections.defaultdict(list)

habitat-baselines/habitat_baselines/rl/ver/queue.py

@@ -50,7 +50,6 @@ def get_many(
             return msgs
 
         def put_many(self, xs, block=True, timeout=10.0):
-
             t_start = time.perf_counter()
             n_put = 0
             for x in xs:

habitat-baselines/habitat_baselines/rl/ver/ver_trainer.py

@@ -383,7 +383,6 @@ def _update_agent(self):
         ppo_cfg = self.config.habitat_baselines.rl.ppo
 
         with self.timer.avg_time("learn"):
-
             t_compute_returns = time.perf_counter()
 
             with self.timer.avg_time("compute returns"), inference_mode():

habitat-lab/habitat/core/dataset.py

@@ -194,7 +194,6 @@ def get_episode_iterator(self, *args: Any, **kwargs: Any) -> Iterator[T]:
         return EpisodeIterator(self.episodes, *args, **kwargs)
 
     def to_json(self) -> str:
-
         result = DatasetJSONEncoder().encode(self)
         return result
 

habitat-lab/habitat/core/embodied_task.py

@@ -261,7 +261,6 @@ def __init__(
         self._is_episode_active = False
 
     def _init_entities(self, entities_configs, register_func) -> OrderedDict:
-
         entities = OrderedDict()
         for entity_name, entity_cfg in entities_configs.items():
             entity_cfg = OmegaConf.create(entity_cfg)

habitat-lab/habitat/core/utils.py

@@ -143,7 +143,6 @@ class DatasetFloatJSONEncoder(DatasetJSONEncoder):
     # Overriding method to inject own `_repr` function for floats with needed
     # precision.
     def iterencode(self, o, _one_shot=False):
-
         markers: Optional[Dict] = {} if self.check_circular else None
         if self.ensure_ascii:
             _encoder = encode_basestring_ascii

habitat-lab/habitat/datasets/vln/r2r_vln_dataset.py

@@ -54,7 +54,6 @@ def __init__(self, config: Optional["DictConfig"] = None) -> None:
     def from_json(
         self, json_str: str, scenes_dir: Optional[str] = None
     ) -> None:
-
         deserialized = json.loads(json_str)
         self.instruction_vocab = VocabDict(
             word_list=deserialized["instruction_vocab"]["word_list"]

habitat-lab/habitat/robots/manipulator.py

@@ -181,7 +181,8 @@ def update(self) -> None:
 
     def reset(self) -> None:
         """Reset the joints on the existing robot.
-        NOTE: only arm and gripper joint motors (not gains) are reset by default, derived class should handle any other changes."""
+        NOTE: only arm and gripper joint motors (not gains) are reset by default, derived class should handle any other changes.
+        """
         self.sim_obj.clear_joint_states()
         self.arm_joint_pos = self.params.arm_init_params
         self._fix_joint_values = None

habitat-lab/habitat/robots/mobile_manipulator.py

@@ -152,6 +152,7 @@ def update(self) -> None:
 
     def reset(self) -> None:
         """Reset the joints on the existing robot.
-        NOTE: only arm and gripper joint motors (not gains) are reset by default, derived class should handle any other changes."""
+        NOTE: only arm and gripper joint motors (not gains) are reset by default, derived class should handle any other changes.
+        """
         Manipulator.reset(self)
         RobotBase.reset(self)

habitat-lab/habitat/robots/robot_base.py

@@ -226,7 +226,6 @@ def is_base_link(self, link_id: int) -> bool:
         )
 
     def update_base(self, rigid_state, target_rigid_state):
-
         end_pos = self._sim.step_filter(
             rigid_state.translation, target_rigid_state.translation
         )

habitat-lab/habitat/robots/static_manipulator.py

@@ -96,5 +96,6 @@ def update(self) -> None:
 
     def reset(self) -> None:
         """Reset the joints on the existing robot.
-        NOTE: only arm and gripper joint motors (not gains) are reset by default, derived class should handle any other changes."""
+        NOTE: only arm and gripper joint motors (not gains) are reset by default, derived class should handle any other changes.
+        """
         Manipulator.reset(self)

habitat-lab/habitat/tasks/nav/instance_image_nav_task.py

@@ -200,7 +200,6 @@ def get_observation(
         episode: InstanceImageGoalNavEpisode,
         **kwargs: Any,
     ) -> Optional[VisualObservation]:
-
         if len(episode.goals) == 0:
             logger.error(
                 f"No goal specified for episode {episode.episode_id}."
@@ -243,7 +242,6 @@ def _get_sensor_type(self, *args: Any, **kwargs: Any):
     def get_observation(
         self, *args: Any, episode: InstanceImageGoalNavEpisode, **kwargs: Any
     ) -> np.ndarray:
-
         if len(episode.goals) == 0:
             logger.error(
                 f"No goal specified for episode {episode.episode_id}."

habitat-lab/habitat/tasks/nav/nav.py

@@ -752,7 +752,6 @@ def _draw_goals_view_points(self, episode):
 
     def _draw_goals_positions(self, episode):
         if self._config.draw_goal_positions:
-
             for goal in episode.goals:
                 if self._is_on_same_floor(goal.position[1]):
                     try:

habitat-lab/habitat/tasks/nav/object_nav_task.py

@@ -152,7 +152,6 @@ def get_observation(
         episode: ObjectGoalNavEpisode,
         **kwargs: Any,
     ) -> Optional[np.ndarray]:
-
         if len(episode.goals) == 0:
             logger.error(
                 f"No goal specified for episode {episode.episode_id}."

habitat-lab/habitat/tasks/nav/shortest_path_follower.py

@@ -44,7 +44,6 @@ def __init__(
         return_one_hot: bool = True,
         stop_on_error: bool = True,
     ):
-
         self._return_one_hot = return_one_hot
         self._sim = sim
         self._goal_radius = goal_radius

habitat-lab/habitat/tasks/rearrange/actions/actions.py

@@ -329,7 +329,6 @@ def _set_robot_state(self, set_dat):
         self.cur_robot.sim_obj.joint_forces = set_dat["pos"]
 
     def update_base(self):
-
         ctrl_freq = self._sim.ctrl_freq
 
         before_trans_state = self._capture_robot_state()

habitat-lab/habitat/tasks/rearrange/multi_task/pddl_action.py

@@ -193,7 +193,6 @@ def init_task(
         should_reset: bool = True,
         add_task_kwargs: Optional[Dict[str, Any]] = None,
     ) -> RearrangeTask:
-
         rearrange_logger.debug(
             f"Loading task {self._task_info.task} with definition {self._task_info.task_def}"
         )

habitat-lab/habitat/tasks/rearrange/sub_tasks/nav_to_obj_task.py

@@ -83,7 +83,6 @@ def _generate_nav_start_goal(self, episode, force_idx=None) -> NavToInfo:
             # Select an object at random and navigate to that object.
             all_pos = self._sim.get_target_objs_start()
             if force_idx is None:
-
                 nav_to_pos = all_pos[np.random.randint(0, len(all_pos))]
             else:
                 nav_to_pos = all_pos[force_idx]

test/test_baselines_hydra.py

@@ -24,7 +24,6 @@ def my_app_compose_api() -> "DictConfig":
     # This is needed for apps that cannot have
     # a standard @hydra.main() entry point
     with initialize(version_base=None):
-
         cfg = compose(
             overrides=[
                 "+habitat_baselines=habitat_baselines_rl_config_base",

test/test_habitat_sim.py

@@ -29,7 +29,6 @@ def test_sim_trajectory():
     with open("test/data/habitat-sim_trajectory_data.json", "r") as f:
         test_trajectory = json.load(f)
     with init_sim() as sim:
-
         sim.reset()
         sim.set_agent_state(
             position=test_trajectory["positions"][0],

test/test_mp3d_eqa.py

@@ -103,7 +103,6 @@ def test_mp3d_eqa_dataset():
 
 @pytest.mark.parametrize("split", ["train", "val"])
 def test_dataset_splitting(split):
-
     dataset_config = get_config(CFG_TEST).habitat.dataset
     with habitat.config.read_write(dataset_config):
         dataset_config.split = split

test/test_pyrobot.py

@@ -53,7 +53,6 @@ def test_pyrobot(mocker):
         _try_register_pyrobot()
 
     with make_sim("PyRobot-v0", config=PyrobotConfig()) as reality:
-
         _ = reality.reset()
         _ = reality.step(
             "go_to_relative",