suppress errors in ax - batch 1

Differential Revision: D23969775

fbshipit-source-id: cb422b648033d3a5932ac6fb72a2d881bd3caa4a

ax/benchmark/botorch_methods.py

@@ -120,7 +120,13 @@ def make_basic_generation_strategy(
             num_trials=-1,
             model_kwargs={
                 "model_constructor": fixed_noise_gp_model_constructor,
-                "transforms": [Winsorize] + Cont_X_trans + Y_trans,  # pyre-ignore[6]
+                # pyre-fixme[58]: `+` is not supported for operand types
+                #  `List[typing.Type[Winsorize]]` and
+                #  `List[typing.Type[ax.modelbridge.transforms.base.Transform]]`.
+                # pyre-fixme[58]: `+` is not supported for operand types
+                #  `List[typing.Type[Winsorize]]` and
+                #  `List[typing.Type[ax.modelbridge.transforms.base.Transform]]`.
+                "transforms": [Winsorize] + Cont_X_trans + Y_trans,
                 "transform_configs": {
                     "Winsorize": {
                         f"winsorization_{t}": v

ax/core/base_trial.py

@@ -398,7 +398,6 @@ def _set_generation_step_index(self, generation_step_index: Optional[int]) -> No
                 "Cannot add generator runs from different generation steps to a "
                 "single trial."
             )
-        # pyre-fixme[8]: Attribute has type `None`; used as `Optional[int]`.
         self._generation_step_index = generation_step_index
 
     @abstractproperty
@@ -601,7 +600,6 @@ def _status(self) -> TrialStatus:
         that the trial statuses mapping on the experiment is updated always when
         a trial status is updated.
         """
-        # pyre-fixme[7]: Expected `TrialStatus` but got `None`.
         return self.__status
 
     @_status.setter
@@ -613,5 +611,4 @@ def _status(self, trial_status: TrialStatus) -> None:
             assert self.index in self._experiment._trial_indices_by_status[self._status]
             self._experiment._trial_indices_by_status[self._status].remove(self.index)
         self._experiment._trial_indices_by_status[trial_status].add(self.index)
-        # pyre-fixme[8]: Attribute has type `None`; used as `TrialStatus`.
         self.__status = trial_status

ax/core/experiment.py

@@ -269,7 +269,6 @@ def optimization_config(self, optimization_config: OptimizationConfig) -> None:
         for metric_name in optimization_config.metrics.keys():
             if metric_name in self._tracking_metrics:
                 self.remove_tracking_metric(metric_name)
-        # pyre-fixme[8]: Attribute has type `None`; used as `OptimizationConfig`.
         self._optimization_config = optimization_config
 
     @property

ax/core/parameter.py

@@ -153,11 +153,11 @@ def _validate_range_param(
             ParameterType.FLOAT,
         ):
             raise ValueError("RangeParameter type must be int or float.")
-        # pyre-fixme[6]: `>=` is not supported for operand types `Union[None, bool,
+        # pyre-fixme[58]: `>=` is not supported for operand types `Union[None, bool,
         #  float, int, str]` and `Union[None, bool, float, int, str]`.
         if lower >= upper:
             raise ValueError("max must be strictly larger than min.")
-        # pyre-fixme[6]: `<=` is not supported for operand types `Union[None, bool,
+        # pyre-fixme[58]: `<=` is not supported for operand types `Union[None, bool,
         #  float, int, str]` and `int`.
         if log_scale and lower <= 0:
             raise ValueError("Cannot take log when min <= 0.")
@@ -236,7 +236,7 @@ def set_digits(self, digits: int) -> "RangeParameter":
         # Re-scale min and max to new digits definition
         cast_lower = self.cast(self._lower)
         cast_upper = self.cast(self._upper)
-        # pyre-fixme[6]: `>=` is not supported for operand types `Union[None, bool,
+        # pyre-fixme[58]: `>=` is not supported for operand types `Union[None, bool,
         #  float, int, str]` and `Union[None, bool, float, int, str]`.
         if cast_lower >= cast_upper:
             raise ValueError(

ax/modelbridge/cross_validation.py

@@ -144,8 +144,8 @@ def cross_validate_by_trial(model: ModelBridge, trial: int = -1) -> List[CVResul
     for obs in training_data:
         if obs.features.trial_index is None:
             continue
-        # pyre-fixme[6]: `<` is not supported for operand types `Optional[np.int64]`
-        #  and `int`.
+        # pyre-fixme[58]: `<` is not supported for operand types
+        #  `Optional[np.int64]` and `int`.
         elif obs.features.trial_index < trial:
             cv_training_data.append(obs)
         elif obs.features.trial_index == trial:

ax/modelbridge/generation_strategy.py

@@ -277,8 +277,6 @@ def trial_indices_by_step(self) -> Dict[int, Set[int]]:
         for trial_index, trial in self.experiment.trials.items():
             if (
                 trial._generation_step_index is not None
-                # pyre-fixme[6]: `<=` is not supported for operand types `None` and
-                #  `int`.
                 and trial._generation_step_index <= self._curr.index
             ):
                 trial_indices_by_step[trial._generation_step_index].add(trial_index)
@@ -592,8 +590,6 @@ def _save_seen_trial_indices(self) -> None:
         has seen the data for). Useful when `use_update=True` for a given
         generation step.
         """
-        # pyre-fixme[8]: Attribute has type `None`; used as `Dict[TrialStatus,
-        #  Set[int]]`.
         self._seen_trial_indices_by_status = deepcopy(
             self.experiment.trial_indices_by_status
         )

ax/models/discrete/thompson.py

@@ -59,23 +59,12 @@ def fit(
         parameter_values: List[TParamValueList],
         outcome_names: List[str],
     ) -> None:
-        # pyre-fixme[8]: Attribute has type `None`; used as
-        #  `List[List[typing.Union[None, bool, float, int, str]]]`.
         self.X = self._fit_X(Xs=Xs)
-        # pyre-fixme[8]: Attribute has type `None`; used as `List[List[float]]`.
-        # pyre-fixme[8]: Attribute has type `None`; used as `List[List[float]]`.
         self.Ys, self.Yvars = self._fit_Ys_and_Yvars(
             Ys=Ys, Yvars=Yvars, outcome_names=outcome_names
         )
-        # pyre-fixme[8]: Attribute has type `None`; used as
-        #  `List[Dict[List[typing.Union[None, bool, float, int, str]], Tuple[float,
-        #  float]]]`.
         self.X_to_Ys_and_Yvars = self._fit_X_to_Ys_and_Yvars(
-            # pyre-fixme[6]: Expected `List[List[typing.Union[None, bool, float,
-            #  int, str]]]` for 1st param but got `None`.
-            X=self.X,
-            Ys=self.Ys,
-            Yvars=self.Yvars,
+            X=self.X, Ys=self.Ys, Yvars=self.Yvars
         )
 
     # pyre-fixme[56]: While applying decorator
@@ -95,7 +84,6 @@ def gen(
             raise ValueError("ThompsonSampler requires objective weights.")
 
         arms = self.X
-        # pyre-fixme[6]: Expected `Sized` for 1st param but got `None`.
         k = len(arms)
 
         weights = self._generate_weights(
@@ -105,10 +93,10 @@ def gen(
 
         # Second entry is used for tie-breaking
         weighted_arms = [
-            # pyre-fixme[16]: `None` has no attribute `__getitem__`.
             (weights[i], np.random.random(), arms[i])
             for i in range(k)
-            # pyre-fixme[6]: Expected `float` for 1st param but got `Optional[float]`.
+            # pyre-fixme[58]: `>` is not supported for operand types `float` and
+            #  `Optional[float]`.
             if weights[i] > min_weight
         ]
 
@@ -134,12 +122,10 @@ def gen(
     @copy_doc(DiscreteModel.predict)
     def predict(self, X: List[TParamValueList]) -> Tuple[np.ndarray, np.ndarray]:
         n = len(X)  # number of parameterizations at which to make predictions
-        # pyre-fixme[6]: Expected `Sized` for 1st param but got `None`.
         m = len(self.Ys)  # number of outcomes
         f = np.zeros((n, m))  # array of outcome predictions
         cov = np.zeros((n, m, m))  # array of predictive covariances
         predictX = [self._hash_TParamValueList(x) for x in X]
-        # pyre-fixme[6]: Expected `Iterable[Variable[_T]]` for 1st param but got `None`.
         for i, X_to_Y_and_Yvar in enumerate(self.X_to_Ys_and_Yvars):
             # iterate through outcomes
             for j, x in enumerate(predictX):
@@ -182,7 +168,6 @@ def _generate_weights(
             num_valid_samples = samples.shape[1]
 
         winner_indices = np.argmax(samples, axis=0)  # (num_samples,)
-        # pyre-fixme[6]: Expected `Sized` for 1st param but got `None`.
         winner_counts = np.zeros(len(self.X))  # (k,)
         for index in winner_indices:
             winner_counts[index] += 1
@@ -195,15 +180,11 @@ def _produce_samples(
         objective_weights: np.ndarray,
         outcome_constraints: Optional[Tuple[np.ndarray, np.ndarray]],
     ) -> Tuple[np.ndarray, float]:
-        # pyre-fixme[6]: Expected `Sized` for 1st param but got `None`.
         k = len(self.X)
         samples_per_metric = np.zeros(
-            # pyre-fixme[6]: Expected `Sized` for 1st param but got `None`.
             (k, num_samples, len(self.Ys))
         )  # k x num_samples x m
-        # pyre-fixme[6]: Expected `Iterable[Variable[_T]]` for 1st param but got `None`.
         for i, Y in enumerate(self.Ys):  # (k x 1)
-            # pyre-fixme[16]: `None` has no attribute `__getitem__`.
             Yvar = self.Yvars[i]  # (k x 1)
             cov = np.diag(Yvar)  # (k x k)
             samples = np.random.multivariate_normal(

ax/models/random/sobol.py

@@ -67,7 +67,6 @@ def init_engine(self, n_tunable_features: int) -> SobolEngine:
             self._engine = SobolEngine(
                 dimension=n_tunable_features, scramble=self.scramble, seed=self.seed
             ).fast_forward(self.init_position)
-        # pyre-fixme[7]: Expected `SobolEngine` but got `None`.
         return self._engine
 
     @property

ax/models/torch/alebo.py

@@ -267,10 +267,8 @@ def get_map_model(
         mll.train()
         mll, info_dict = fit_gpytorch_scipy(mll, track_iterations=False, method="tnc")
         logger.debug(info_dict)
-        # pyre-fixme[6]: Expected `List[botorch.optim.fit.OptimizationIteration]`
-        #  for 1st param but got `float`.
-        # pyre-fixme[6]: Expected `List[botorch.optim.fit.OptimizationIteration]`
-        #  for 1st param but got `float`.
+        # pyre-fixme[58]: `<` is not supported for operand types
+        #  `Union[List[botorch.optim.fit.OptimizationIteration], float]` and `float`.
         if info_dict["fopt"] < f_best:
             f_best = float(info_dict["fopt"])  # pyre-ignore
             sd_best = m.state_dict()

ax/plot/table_view.py

@@ -144,7 +144,8 @@ def transpose(m):
 
     records = [[name.replace(":", " : ") for name in metric_names]] + transpose(records)
     colors = [["#ffffff"] * len(metric_names)] + transpose(colors)
-    # pyre-fixme[6]: Expected `List[str]` for 1st param but got `List[float]`.
+    # pyre-fixme[58]: `+` is not supported for operand types `List[str]` and
+    #  `List[float]`.
     header = [f"<b>{x}</b>" for x in [f"{arm_noun}s"] + arm_names]
     column_widths = [300] + [150] * len(arm_names)
 

ax/storage/json_store/decoders.py

@@ -71,7 +71,6 @@ def batch_trial_from_json(
     batch._status_quo = status_quo
     batch._status_quo_weight_override = status_quo_weight_override
     batch.optimize_for_power = optimize_for_power
-    # pyre-fixme[8]: Attribute has type `None`; used as `Optional[int]`.
     batch._generation_step_index = generation_step_index
     batch._properties = properties
     return batch
@@ -120,7 +119,6 @@ def trial_from_json(
     trial._run_metadata = run_metadata or {}
     trial._runner = runner
     trial._num_arms_created = num_arms_created
-    # pyre-fixme[8]: Attribute has type `None`; used as `Optional[int]`.
     trial._generation_step_index = generation_step_index
     trial._properties = properties or {}
     return trial

ax/storage/sqa_store/decoder.py

@@ -751,8 +751,6 @@ def trial_from_sqa(self, trial_sqa: SQATrial, experiment: Experiment) -> BaseTri
         trial._runner = (
             self.runner_from_sqa(trial_sqa.runner) if trial_sqa.runner else None
         )
-        # pyre-fixme[9]: _generation_step_index has type `None`; used as
-        #  `Optional[int]`.
         trial._generation_step_index = trial_sqa.generation_step_index
         trial._properties = trial_sqa.properties or {}
         return trial

ax/utils/measurement/synthetic_functions.py

@@ -119,7 +119,6 @@ def domain(self) -> List[Tuple[float, ...]]:
         where each element of the list is a tuple corresponding to the min
         and max of the domain for that dimension.
         """
-        # pyre-fixme[7]: Expected `List[typing.Tuple[float, ...]]` but got `None`.
         return self._domain
 
     @property
@@ -130,7 +129,6 @@ def minimums(self) -> List[Tuple[float, ...]]:
         Each element of the list is a d-tuple, where d is the dimensionality
         of the inputs. There may be more than one global minimums.
         """
-        # pyre-fixme[7]: Expected `List[typing.Tuple[float, ...]]` but got `None`.
         return self._minimums
 
     @property
@@ -141,21 +139,18 @@ def maximums(self) -> List[Tuple[float, ...]]:
         Each element of the list is a d-tuple, where d is the dimensionality
         of the inputs. There may be more than one global minimums.
         """
-        # pyre-fixme[7]: Expected `List[typing.Tuple[float, ...]]` but got `None`.
         return self._maximums
 
     @property
     @informative_failure_on_none
     def fmin(self) -> float:
         """Value at global minimum(s)."""
-        # pyre-fixme[7]: Expected `float` but got `None`.
         return self._fmin
 
     @property
     @informative_failure_on_none
     def fmax(self) -> float:
         """Value at global minimum(s)."""
-        # pyre-fixme[7]: Expected `float` but got `None`.
         return self._fmax
 
     @classmethod
@@ -202,16 +197,10 @@ def from_botorch(
 class Hartmann6(SyntheticFunction):
     """Hartmann6 function (6-dimensional with 1 global minimum)."""
 
-    # pyre-fixme[15]: `_required_dimensionality` overrides attribute defined in
-    #  `SyntheticFunction` inconsistently.
     _required_dimensionality = 6
     _domain = [(0, 1) for i in range(6)]
-    # pyre-fixme[15]: `_minimums` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _minimums = [(0.20169, 0.150011, 0.476874, 0.275332, 0.311652, 0.6573)]
     _fmin = -3.32237
-    # pyre-fixme[15]: `_fmax` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _fmax = 0.0
     _alpha = np.array([1.0, 1.2, 3.0, 3.2])
     _A = np.array(
@@ -277,18 +266,10 @@ def _f(self, X: np.ndarray) -> float:
 class Branin(SyntheticFunction):
     """Branin function (2-dimensional with 3 global minima)."""
 
-    # pyre-fixme[15]: `_required_dimensionality` overrides attribute defined in
-    #  `SyntheticFunction` inconsistently.
     _required_dimensionality = 2
-    # pyre-fixme[15]: `_domain` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _domain = [(-5, 10), (0, 15)]
     _minimums = [(-np.pi, 12.275), (np.pi, 2.275), (9.42478, 2.475)]
-    # pyre-fixme[15]: `_fmin` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _fmin = 0.397887
-    # pyre-fixme[15]: `_fmax` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _fmax = 294.0
 
     # pyre-fixme[56]: While applying decorator
@@ -308,18 +289,10 @@ def _f(self, X: np.ndarray) -> float:
 class Aug_Branin(SyntheticFunction):
     """Augmented Branin function (3-dimensional with infinitely many global minima)."""
 
-    # pyre-fixme[15]: `_required_dimensionality` overrides attribute defined in
-    #  `SyntheticFunction` inconsistently.
     _required_dimensionality = 3
-    # pyre-fixme[15]: `_domain` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _domain = [(-5, 10), (0, 15), (0, 1)]
     _minimums = [(-np.pi, 12.275, 1), (np.pi, 2.275, 1), (9.42478, 2.475, 1)]
-    # pyre-fixme[15]: `_fmin` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _fmin = 0.397887
-    # pyre-fixme[15]: `_fmax` overrides attribute defined in `SyntheticFunction`
-    #  inconsistently.
     _fmax = 294.0
 
     # pyre-fixme[56]: While applying decorator

ax/utils/testing/core_stubs.py

@@ -459,7 +459,6 @@ def get_batch_trial(abandon_arm: bool = True) -> BatchTrial:
         batch.mark_arm_abandoned(batch.arms[0].name, "abandoned reason")
     batch.runner = SyntheticRunner()
     batch.set_status_quo_with_weight(status_quo=arms[0], weight=0.5)
-    # pyre-fixme[8]: Attribute has type `None`; used as `int`.
     batch._generation_step_index = 0
     return batch
 
@@ -513,7 +512,6 @@ def get_trial() -> Trial:
     arm = get_arms_from_dict(get_arm_weights1())[0]
     trial.add_arm(arm)
     trial.runner = SyntheticRunner()
-    # pyre-fixme[8]: Attribute has type `None`; used as `int`.
     trial._generation_step_index = 0
     return trial
 

ax/utils/testing/modeling_stubs.py

@@ -172,7 +172,7 @@ def transform_optimization_config(
         fixed_features: ObservationFeatures,
     ) -> OptimizationConfig:
         return (  # pyre-ignore[7]: pyre is right, this is a hack for testing.
-            # pyre-fixme[6]: `+` is not supported for operand types
+            # pyre-fixme[58]: `+` is not supported for operand types
             #  `OptimizationConfig` and `int`.
             optimization_config + 1
             if isinstance(optimization_config, int)
@@ -185,7 +185,10 @@ def transform_observation_features(
         for obsf in observation_features:
             if "x" in obsf.parameters:
                 obsf.parameters["x"] = (
-                    not_none(obsf.parameters["x"]) + 1  # pyre-ignore[6]
+                    # pyre-fixme[58]: `+` is not supported for operand types
+                    #  `Union[float, str]` and `int`.
+                    not_none(obsf.parameters["x"])
+                    + 1
                 )
         return observation_features
 
@@ -228,7 +231,7 @@ def transform_optimization_config(
         fixed_features: ObservationFeatures,
     ) -> OptimizationConfig:
         return (
-            # pyre-fixme[6]: `**` is not supported for operand types
+            # pyre-fixme[58]: `**` is not supported for operand types
             #  `OptimizationConfig` and `int`.
             optimization_config ** 2
             if isinstance(optimization_config, int)