mirror of
https://github.com/huggingface/lerobot.git
synced 2026-05-15 16:49:55 +00:00
chore(processor): add Step suffix to all processors (#1854)
* refactor(processor): rename MapDeltaActionToRobotAction and MapTensorToDeltaActionDict for consistency * refactor(processor): rename DeviceProcessor to DeviceProcessorStep for consistency across modules * refactor(processor): rename Torch2NumpyActionProcessor to Torch2NumpyActionProcessorStep for consistency * refactor(processor): rename Numpy2TorchActionProcessor to Numpy2TorchActionProcessorStep for consistency * refactor(processor): rename AddTeleopActionAsComplimentaryData to AddTeleopActionAsComplimentaryDataStep for consistency * refactor(processor): rename ImageCropResizeProcessor and AddTeleopEventsAsInfo for consistency * refactor(processor): rename TimeLimitProcessor to TimeLimitProcessorStep for consistency * refactor(processor): rename GripperPenaltyProcessor to GripperPenaltyProcessorStep for consistency * refactor(processor): rename InterventionActionProcessor to InterventionActionProcessorStep for consistency * refactor(processor): rename RewardClassifierProcessor to RewardClassifierProcessorStep for consistency * refactor(processor): rename JointVelocityProcessor to JointVelocityProcessorStep for consistency * refactor(processor): rename MotorCurrentProcessor to MotorCurrentProcessorStep for consistency * refactor(processor): rename NormalizerProcessor and UnnormalizerProcessor to NormalizerProcessorStep and UnnormalizerProcessorStep for consistency * refactor(processor): rename VanillaObservationProcessor to VanillaObservationProcessorStep for consistency * refactor(processor): rename RenameProcessor to RenameProcessorStep for consistency * refactor(processor): rename TokenizerProcessor to TokenizerProcessorStep for consistency * refactor(processor): rename ToBatchProcessor to AddBatchDimensionProcessorStep for consistency * refactor(processor): update config file name in test for RenameProcessorStep consistency
This commit is contained in:
Adil Zouitine
@@ -23,9 +23,9 @@ from lerobot.configs.types import FeatureType, NormalizationMode, PolicyFeature
|
||||
from lerobot.processor import (
|
||||
DataProcessorPipeline,
|
||||
IdentityProcessorStep,
|
||||
NormalizerProcessor,
|
||||
NormalizerProcessorStep,
|
||||
TransitionKey,
|
||||
UnnormalizerProcessor,
|
||||
UnnormalizerProcessorStep,
|
||||
hotswap_stats,
|
||||
)
|
||||
from lerobot.processor.converters import to_tensor
|
||||
@@ -125,7 +125,7 @@ def _create_observation_norm_map():
|
||||
}
|
||||
|
||||
|
||||
# Fixtures for observation normalisation tests using NormalizerProcessor
|
||||
# Fixtures for observation normalisation tests using NormalizerProcessorStep
|
||||
@pytest.fixture
|
||||
def observation_stats():
|
||||
return {
|
||||
@@ -142,10 +142,10 @@ def observation_stats():
|
||||
|
||||
@pytest.fixture
|
||||
def observation_normalizer(observation_stats):
|
||||
"""Return a NormalizerProcessor that only has observation stats (no action)."""
|
||||
"""Return a NormalizerProcessorStep that only has observation stats (no action)."""
|
||||
features = _create_observation_features()
|
||||
norm_map = _create_observation_norm_map()
|
||||
return NormalizerProcessor(features=features, norm_map=norm_map, stats=observation_stats)
|
||||
return NormalizerProcessorStep(features=features, norm_map=norm_map, stats=observation_stats)
|
||||
|
||||
|
||||
def test_mean_std_normalization(observation_normalizer):
|
||||
@@ -182,7 +182,7 @@ def test_min_max_normalization(observation_normalizer):
|
||||
def test_selective_normalization(observation_stats):
|
||||
features = _create_observation_features()
|
||||
norm_map = _create_observation_norm_map()
|
||||
normalizer = NormalizerProcessor(
|
||||
normalizer = NormalizerProcessorStep(
|
||||
features=features,
|
||||
norm_map=norm_map,
|
||||
stats=observation_stats,
|
||||
@@ -208,7 +208,7 @@ def test_selective_normalization(observation_stats):
|
||||
def test_device_compatibility(observation_stats):
|
||||
features = _create_observation_features()
|
||||
norm_map = _create_observation_norm_map()
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=observation_stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=observation_stats)
|
||||
observation = {
|
||||
"observation.image": torch.tensor([0.7, 0.5, 0.3]).cuda(),
|
||||
}
|
||||
@@ -237,7 +237,7 @@ def test_from_lerobot_dataset():
|
||||
FeatureType.ACTION: NormalizationMode.MEAN_STD,
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor.from_lerobot_dataset(mock_dataset, features, norm_map)
|
||||
normalizer = NormalizerProcessorStep.from_lerobot_dataset(mock_dataset, features, norm_map)
|
||||
|
||||
# Both observation and action statistics should be present in tensor stats
|
||||
assert "observation.image" in normalizer._tensor_stats
|
||||
@@ -252,7 +252,7 @@ def test_state_dict_save_load(observation_normalizer):
|
||||
# Create new normalizer and load state
|
||||
features = _create_observation_features()
|
||||
norm_map = _create_observation_norm_map()
|
||||
new_normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats={})
|
||||
new_normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats={})
|
||||
new_normalizer.load_state_dict(state_dict)
|
||||
|
||||
# Test that it works the same
|
||||
@@ -303,7 +303,7 @@ def _create_action_norm_map_min_max():
|
||||
def test_mean_std_unnormalization(action_stats_mean_std):
|
||||
features = _create_action_features()
|
||||
norm_map = _create_action_norm_map_mean_std()
|
||||
unnormalizer = UnnormalizerProcessor(
|
||||
unnormalizer = UnnormalizerProcessorStep(
|
||||
features=features, norm_map=norm_map, stats={"action": action_stats_mean_std}
|
||||
)
|
||||
|
||||
@@ -321,7 +321,7 @@ def test_mean_std_unnormalization(action_stats_mean_std):
|
||||
def test_min_max_unnormalization(action_stats_min_max):
|
||||
features = _create_action_features()
|
||||
norm_map = _create_action_norm_map_min_max()
|
||||
unnormalizer = UnnormalizerProcessor(
|
||||
unnormalizer = UnnormalizerProcessorStep(
|
||||
features=features, norm_map=norm_map, stats={"action": action_stats_min_max}
|
||||
)
|
||||
|
||||
@@ -347,7 +347,7 @@ def test_min_max_unnormalization(action_stats_min_max):
|
||||
def test_numpy_action_input(action_stats_mean_std):
|
||||
features = _create_action_features()
|
||||
norm_map = _create_action_norm_map_mean_std()
|
||||
unnormalizer = UnnormalizerProcessor(
|
||||
unnormalizer = UnnormalizerProcessorStep(
|
||||
features=features, norm_map=norm_map, stats={"action": action_stats_mean_std}
|
||||
)
|
||||
|
||||
@@ -365,7 +365,7 @@ def test_numpy_action_input(action_stats_mean_std):
|
||||
def test_none_action(action_stats_mean_std):
|
||||
features = _create_action_features()
|
||||
norm_map = _create_action_norm_map_mean_std()
|
||||
unnormalizer = UnnormalizerProcessor(
|
||||
unnormalizer = UnnormalizerProcessorStep(
|
||||
features=features, norm_map=norm_map, stats={"action": action_stats_mean_std}
|
||||
)
|
||||
|
||||
@@ -381,11 +381,11 @@ def test_action_from_lerobot_dataset():
|
||||
mock_dataset.meta.stats = {"action": {"mean": [0.0], "std": [1.0]}}
|
||||
features = {"action": PolicyFeature(FeatureType.ACTION, (1,))}
|
||||
norm_map = {FeatureType.ACTION: NormalizationMode.MEAN_STD}
|
||||
unnormalizer = UnnormalizerProcessor.from_lerobot_dataset(mock_dataset, features, norm_map)
|
||||
unnormalizer = UnnormalizerProcessorStep.from_lerobot_dataset(mock_dataset, features, norm_map)
|
||||
assert "mean" in unnormalizer._tensor_stats["action"]
|
||||
|
||||
|
||||
# Fixtures for NormalizerProcessor tests
|
||||
# Fixtures for NormalizerProcessorStep tests
|
||||
@pytest.fixture
|
||||
def full_stats():
|
||||
return {
|
||||
@@ -424,7 +424,7 @@ def _create_full_norm_map():
|
||||
def normalizer_processor(full_stats):
|
||||
features = _create_full_features()
|
||||
norm_map = _create_full_norm_map()
|
||||
return NormalizerProcessor(features=features, norm_map=norm_map, stats=full_stats)
|
||||
return NormalizerProcessorStep(features=features, norm_map=norm_map, stats=full_stats)
|
||||
|
||||
|
||||
def test_combined_normalization(normalizer_processor):
|
||||
@@ -468,7 +468,7 @@ def test_processor_from_lerobot_dataset(full_stats):
|
||||
features = _create_full_features()
|
||||
norm_map = _create_full_norm_map()
|
||||
|
||||
processor = NormalizerProcessor.from_lerobot_dataset(
|
||||
processor = NormalizerProcessorStep.from_lerobot_dataset(
|
||||
mock_dataset, features, norm_map, normalize_observation_keys={"observation.image"}
|
||||
)
|
||||
|
||||
@@ -480,7 +480,7 @@ def test_processor_from_lerobot_dataset(full_stats):
|
||||
def test_get_config(full_stats):
|
||||
features = _create_full_features()
|
||||
norm_map = _create_full_norm_map()
|
||||
processor = NormalizerProcessor(
|
||||
processor = NormalizerProcessorStep(
|
||||
features=features,
|
||||
norm_map=norm_map,
|
||||
stats=full_stats,
|
||||
@@ -539,7 +539,7 @@ def test_empty_observation():
|
||||
stats = {"observation.image": {"mean": [0.5], "std": [0.2]}}
|
||||
features = {"observation.image": PolicyFeature(FeatureType.VISUAL, (3, 96, 96))}
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
transition = create_transition()
|
||||
result = normalizer(transition)
|
||||
@@ -550,7 +550,7 @@ def test_empty_observation():
|
||||
def test_empty_stats():
|
||||
features = {"observation.image": PolicyFeature(FeatureType.VISUAL, (3, 96, 96))}
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats={})
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats={})
|
||||
observation = {"observation.image": torch.tensor([0.5])}
|
||||
transition = create_transition(observation=observation)
|
||||
|
||||
@@ -566,7 +566,7 @@ def test_partial_stats():
|
||||
stats = {"observation.image": {"mean": [0.5]}} # Missing std / (min,max)
|
||||
features = {"observation.image": PolicyFeature(FeatureType.VISUAL, (3, 96, 96))}
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
observation = {"observation.image": torch.tensor([0.7])}
|
||||
transition = create_transition(observation=observation)
|
||||
|
||||
@@ -581,7 +581,7 @@ def test_missing_action_stats_no_error():
|
||||
features = {"observation.image": PolicyFeature(FeatureType.VISUAL, (3, 96, 96))}
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
|
||||
processor = UnnormalizerProcessor.from_lerobot_dataset(mock_dataset, features, norm_map)
|
||||
processor = UnnormalizerProcessorStep.from_lerobot_dataset(mock_dataset, features, norm_map)
|
||||
# The tensor stats should not contain the 'action' key
|
||||
assert "action" not in processor._tensor_stats
|
||||
|
||||
@@ -590,7 +590,7 @@ def test_serialization_roundtrip(full_stats):
|
||||
"""Test that features and norm_map can be serialized and deserialized correctly."""
|
||||
features = _create_full_features()
|
||||
norm_map = _create_full_norm_map()
|
||||
original_processor = NormalizerProcessor(
|
||||
original_processor = NormalizerProcessorStep(
|
||||
features=features,
|
||||
norm_map=norm_map,
|
||||
stats=full_stats,
|
||||
@@ -602,7 +602,7 @@ def test_serialization_roundtrip(full_stats):
|
||||
config = original_processor.get_config()
|
||||
|
||||
# Create a new processor from the config (deserialization)
|
||||
new_processor = NormalizerProcessor(
|
||||
new_processor = NormalizerProcessorStep(
|
||||
features=config["features"],
|
||||
norm_map=config["norm_map"],
|
||||
stats=full_stats,
|
||||
@@ -670,7 +670,7 @@ def test_identity_normalization_observations():
|
||||
"observation.state": {"mean": [0.0, 0.0], "std": [1.0, 1.0]},
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
observation = {
|
||||
"observation.image": torch.tensor([0.7, 0.5, 0.3]),
|
||||
@@ -695,7 +695,7 @@ def test_identity_normalization_actions():
|
||||
norm_map = {FeatureType.ACTION: NormalizationMode.IDENTITY}
|
||||
stats = {"action": {"mean": [0.0, 0.0], "std": [1.0, 2.0]}}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
action = torch.tensor([1.0, -0.5])
|
||||
transition = create_transition(action=action)
|
||||
@@ -721,7 +721,7 @@ def test_identity_unnormalization_observations():
|
||||
"observation.state": {"min": [-1.0, -1.0], "max": [1.0, 1.0]},
|
||||
}
|
||||
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
observation = {
|
||||
"observation.image": torch.tensor([0.7, 0.5, 0.3]),
|
||||
@@ -748,7 +748,7 @@ def test_identity_unnormalization_actions():
|
||||
norm_map = {FeatureType.ACTION: NormalizationMode.IDENTITY}
|
||||
stats = {"action": {"min": [-1.0, -2.0], "max": [1.0, 2.0]}}
|
||||
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
action = torch.tensor([0.5, -0.8]) # Normalized values
|
||||
transition = create_transition(action=action)
|
||||
@@ -771,8 +771,8 @@ def test_identity_with_missing_stats():
|
||||
}
|
||||
stats = {} # No stats provided
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
observation = {"observation.image": torch.tensor([0.7, 0.5, 0.3])}
|
||||
action = torch.tensor([1.0, -0.5])
|
||||
@@ -812,7 +812,7 @@ def test_identity_mixed_with_other_modes():
|
||||
"action": {"min": [-1.0, -1.0], "max": [1.0, 1.0]},
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
observation = {
|
||||
"observation.image": torch.tensor([0.7, 0.5, 0.3]),
|
||||
@@ -854,7 +854,7 @@ def test_identity_defaults_when_not_in_norm_map():
|
||||
"observation.state": {"mean": [0.0, 0.0], "std": [1.0, 1.0]},
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
observation = {
|
||||
"observation.image": torch.tensor([0.7, 0.5, 0.3]),
|
||||
@@ -888,8 +888,8 @@ def test_identity_roundtrip():
|
||||
"action": {"min": [-1.0, -1.0], "max": [1.0, 1.0]},
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
original_observation = {"observation.image": torch.tensor([0.7, 0.5, 0.3])}
|
||||
original_action = torch.tensor([0.5, -0.2])
|
||||
@@ -921,7 +921,7 @@ def test_identity_config_serialization():
|
||||
"action": {"mean": [0.0, 0.0], "std": [1.0, 1.0]},
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
# Get config
|
||||
config = normalizer.get_config()
|
||||
@@ -931,7 +931,7 @@ def test_identity_config_serialization():
|
||||
assert config["norm_map"]["ACTION"] == "MEAN_STD"
|
||||
|
||||
# Create new processor from config (simulating load)
|
||||
new_normalizer = NormalizerProcessor(
|
||||
new_normalizer = NormalizerProcessorStep(
|
||||
features=config["features"],
|
||||
norm_map=config["norm_map"],
|
||||
stats=stats,
|
||||
@@ -969,7 +969,7 @@ def test_identity_config_serialization():
|
||||
# norm_map = {FeatureType.STATE: NormalizationMode.MEAN_STD}
|
||||
# stats = {"observation.state": {"mean": [0.0, 0.0], "std": [1.0, 1.0]}}
|
||||
|
||||
# normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
# normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
# # Manually inject an invalid mode to test error handling
|
||||
# normalizer.norm_map[FeatureType.STATE] = "INVALID_MODE"
|
||||
@@ -1006,8 +1006,8 @@ def test_hotswap_stats_basic_functionality():
|
||||
}
|
||||
|
||||
# Create processors
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
identity = IdentityProcessorStep()
|
||||
|
||||
# Create robot processor
|
||||
@@ -1047,7 +1047,7 @@ def test_hotswap_stats_deep_copy():
|
||||
}
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
original_processor = DataProcessorPipeline(steps=[normalizer])
|
||||
|
||||
# Store reference to original stats
|
||||
@@ -1072,7 +1072,7 @@ def test_hotswap_stats_deep_copy():
|
||||
|
||||
|
||||
def test_hotswap_stats_only_affects_normalizer_steps():
|
||||
"""Test that hotswap_stats only modifies NormalizerProcessor and UnnormalizerProcessor steps."""
|
||||
"""Test that hotswap_stats only modifies NormalizerProcessorStep and UnnormalizerProcessorStep steps."""
|
||||
stats = {
|
||||
"observation.image": {"mean": np.array([0.5]), "std": np.array([0.2])},
|
||||
}
|
||||
@@ -1087,8 +1087,8 @@ def test_hotswap_stats_only_affects_normalizer_steps():
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
|
||||
# Create mixed steps
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
identity = IdentityProcessorStep()
|
||||
|
||||
robot_processor = DataProcessorPipeline(steps=[normalizer, identity, unnormalizer])
|
||||
@@ -1117,7 +1117,7 @@ def test_hotswap_stats_empty_stats():
|
||||
}
|
||||
norm_map = {FeatureType.VISUAL: NormalizationMode.MEAN_STD}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
robot_processor = DataProcessorPipeline(steps=[normalizer])
|
||||
|
||||
# Hotswap with empty stats
|
||||
@@ -1167,7 +1167,7 @@ def test_hotswap_stats_preserves_other_attributes():
|
||||
normalize_observation_keys = {"observation.image"}
|
||||
eps = 1e-6
|
||||
|
||||
normalizer = NormalizerProcessor(
|
||||
normalizer = NormalizerProcessorStep(
|
||||
features=features,
|
||||
norm_map=norm_map,
|
||||
stats=initial_stats,
|
||||
@@ -1212,10 +1212,10 @@ def test_hotswap_stats_multiple_normalizer_types():
|
||||
}
|
||||
|
||||
# Create multiple normalizers and unnormalizers
|
||||
normalizer1 = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer2 = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
unnormalizer1 = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
unnormalizer2 = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer1 = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer2 = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
unnormalizer1 = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
unnormalizer2 = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
|
||||
robot_processor = DataProcessorPipeline(steps=[normalizer1, unnormalizer1, normalizer2, unnormalizer2])
|
||||
|
||||
@@ -1264,7 +1264,7 @@ def test_hotswap_stats_with_different_data_types():
|
||||
FeatureType.ACTION: NormalizationMode.MEAN_STD,
|
||||
}
|
||||
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
robot_processor = DataProcessorPipeline(steps=[normalizer])
|
||||
|
||||
# Hotswap stats
|
||||
@@ -1320,7 +1320,7 @@ def test_hotswap_stats_functional_test():
|
||||
}
|
||||
|
||||
# Create original processor
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=initial_stats)
|
||||
original_processor = DataProcessorPipeline(
|
||||
steps=[normalizer], to_transition=lambda x: x, to_output=lambda x: x
|
||||
)
|
||||
@@ -1366,7 +1366,7 @@ def test_zero_std_uses_eps():
|
||||
features = {"observation.state": PolicyFeature(FeatureType.STATE, (1,))}
|
||||
norm_map = {FeatureType.STATE: NormalizationMode.MEAN_STD}
|
||||
stats = {"observation.state": {"mean": np.array([0.5]), "std": np.array([0.0])}}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats, eps=1e-6)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats, eps=1e-6)
|
||||
|
||||
observation = {"observation.state": torch.tensor([0.5])} # equals mean
|
||||
out = normalizer(create_transition(observation=observation))
|
||||
@@ -1378,7 +1378,7 @@ def test_min_equals_max_maps_to_minus_one():
|
||||
features = {"observation.state": PolicyFeature(FeatureType.STATE, (1,))}
|
||||
norm_map = {FeatureType.STATE: NormalizationMode.MIN_MAX}
|
||||
stats = {"observation.state": {"min": np.array([2.0]), "max": np.array([2.0])}}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats, eps=1e-6)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats, eps=1e-6)
|
||||
|
||||
observation = {"observation.state": torch.tensor([2.0])}
|
||||
out = normalizer(create_transition(observation=observation))
|
||||
@@ -1393,7 +1393,7 @@ def test_action_normalized_despite_normalize_observation_keys():
|
||||
}
|
||||
norm_map = {FeatureType.STATE: NormalizationMode.IDENTITY, FeatureType.ACTION: NormalizationMode.MEAN_STD}
|
||||
stats = {"action": {"mean": np.array([1.0, -1.0]), "std": np.array([2.0, 4.0])}}
|
||||
normalizer = NormalizerProcessor(
|
||||
normalizer = NormalizerProcessorStep(
|
||||
features=features, norm_map=norm_map, stats=stats, normalize_observation_keys={"observation.state"}
|
||||
)
|
||||
|
||||
@@ -1411,12 +1411,12 @@ def test_unnormalize_observations_mean_std_and_min_max():
|
||||
"observation.mm": PolicyFeature(FeatureType.STATE, (2,)),
|
||||
}
|
||||
# Build two processors: one mean/std and one min/max
|
||||
unnorm_ms = UnnormalizerProcessor(
|
||||
unnorm_ms = UnnormalizerProcessorStep(
|
||||
features={"observation.ms": features["observation.ms"]},
|
||||
norm_map={FeatureType.STATE: NormalizationMode.MEAN_STD},
|
||||
stats={"observation.ms": {"mean": np.array([1.0, -1.0]), "std": np.array([2.0, 4.0])}},
|
||||
)
|
||||
unnorm_mm = UnnormalizerProcessor(
|
||||
unnorm_mm = UnnormalizerProcessorStep(
|
||||
features={"observation.mm": features["observation.mm"]},
|
||||
norm_map={FeatureType.STATE: NormalizationMode.MIN_MAX},
|
||||
stats={"observation.mm": {"min": np.array([0.0, -2.0]), "max": np.array([2.0, 2.0])}},
|
||||
@@ -1438,7 +1438,7 @@ def test_unknown_observation_keys_ignored():
|
||||
features = {"observation.state": PolicyFeature(FeatureType.STATE, (1,))}
|
||||
norm_map = {FeatureType.STATE: NormalizationMode.MEAN_STD}
|
||||
stats = {"observation.state": {"mean": np.array([0.0]), "std": np.array([1.0])}}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
obs = {"observation.state": torch.tensor([1.0]), "observation.unknown": torch.tensor([5.0])}
|
||||
tr = create_transition(observation=obs)
|
||||
@@ -1452,7 +1452,7 @@ def test_batched_action_normalization():
|
||||
features = {"action": PolicyFeature(FeatureType.ACTION, (2,))}
|
||||
norm_map = {FeatureType.ACTION: NormalizationMode.MEAN_STD}
|
||||
stats = {"action": {"mean": np.array([1.0, -1.0]), "std": np.array([2.0, 4.0])}}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
actions = torch.tensor([[1.0, -1.0], [3.0, 3.0]]) # first equals mean → zeros; second → [1, 1]
|
||||
out = normalizer(create_transition(action=actions))[TransitionKey.ACTION]
|
||||
@@ -1464,7 +1464,7 @@ def test_complementary_data_preservation():
|
||||
features = {"observation.state": PolicyFeature(FeatureType.STATE, (1,))}
|
||||
norm_map = {FeatureType.STATE: NormalizationMode.MEAN_STD}
|
||||
stats = {"observation.state": {"mean": np.array([0.0]), "std": np.array([1.0])}}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
comp = {"existing": 123}
|
||||
tr = create_transition(observation={"observation.state": torch.tensor([1.0])}, complementary_data=comp)
|
||||
@@ -1483,8 +1483,8 @@ def test_roundtrip_normalize_unnormalize_non_identity():
|
||||
"observation.state": {"mean": np.array([1.0, -1.0]), "std": np.array([2.0, 4.0])},
|
||||
"action": {"min": np.array([-2.0, 0.0]), "max": np.array([2.0, 4.0])},
|
||||
}
|
||||
normalizer = NormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessor(features=features, norm_map=norm_map, stats=stats)
|
||||
normalizer = NormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
unnormalizer = UnnormalizerProcessorStep(features=features, norm_map=norm_map, stats=stats)
|
||||
|
||||
# Add a time dimension in action for broadcasting check (B,T,D)
|
||||
obs = {"observation.state": torch.tensor([[3.0, 3.0], [1.0, -1.0]])}
|
||||
|
||||
Reference in New Issue
Block a user