def load(cls, path: Text) -> Policy:
filename = os.path.join(path, "sklearn_model.pkl")
zero_features_filename = os.path.join(path, "zero_state_features.pkl")
if not os.path.exists(path):
raise OSError("Failed to load dialogue model. Path {} "
"doesn't exist".format(os.path.abspath(filename)))
featurizer = TrackerFeaturizer.load(path)
assert isinstance(featurizer, MaxHistoryTrackerFeaturizer), (
"Loaded featurizer of type {}, should be "
"MaxHistoryTrackerFeaturizer.".format(type(featurizer).__name__))
meta_file = os.path.join(path, "sklearn_policy.json")
meta = json.loads(rasa.shared.utils.io.read_file(meta_file))
zero_state_features = io_utils.pickle_load(zero_features_filename)
policy = cls(
featurizer=featurizer,
priority=meta["priority"],
zero_state_features=zero_state_features,
)
state = io_utils.pickle_load(filename)
vars(policy).update(state)
logger.info("Loaded sklearn model")
return policy
def load(cls, path: Union[Text, Path]) -> Policy:
filename = Path(path) / "sklearn_model.pkl"
zero_features_filename = Path(path) / "zero_state_features.pkl"
if not Path(path).exists():
raise OSError(
f"Failed to load dialogue model. Path {filename.absolute()} "
f"doesn't exist.")
featurizer = TrackerFeaturizer.load(path)
assert isinstance(featurizer, MaxHistoryTrackerFeaturizer), (
f"Loaded featurizer of type {type(featurizer).__name__}, should be "
f"MaxHistoryTrackerFeaturizer.")
meta_file = Path(path) / "sklearn_policy.json"
meta = json.loads(rasa.shared.utils.io.read_file(meta_file))
zero_state_features = io_utils.pickle_load(zero_features_filename)
policy = cls(
featurizer=featurizer,
priority=meta["priority"],
zero_state_features=zero_state_features,
)
state = io_utils.pickle_load(filename)
vars(policy).update(state)
logger.info("Loaded sklearn model")
return policy
def load(
cls,
config: Dict[Text, Any],
model_storage: ModelStorage,
resource: Resource,
execution_context: ExecutionContext,
**kwargs: Any,
) -> Policy:
"""Loads a trained policy (see parent class for full docstring)."""
featurizer = None
try:
with model_storage.read_from(resource) as path:
if (Path(path) / FEATURIZER_FILE).is_file():
featurizer = TrackerFeaturizer.load(path)
config.update(kwargs)
except (ValueError, FileNotFoundError, FileIOException):
logger.debug(
f"Couldn't load metadata for policy '{cls.__name__}' as the persisted "
f"metadata couldn't be loaded."
)
return cls(
config, model_storage, resource, execution_context, featurizer=featurizer,
)
def load(
cls,
config: Dict[Text, Any],
model_storage: ModelStorage,
resource: Resource,
execution_context: ExecutionContext,
**kwargs: Any,
) -> MemoizationPolicy:
"""Loads a trained policy (see parent class for full docstring)."""
featurizer = None
lookup = None
try:
with model_storage.read_from(resource) as path:
metadata_file = Path(path) / cls._metadata_filename()
metadata = rasa.shared.utils.io.read_json_file(metadata_file)
lookup = metadata["lookup"]
if (Path(path) / FEATURIZER_FILE).is_file():
featurizer = TrackerFeaturizer.load(path)
except (ValueError, FileNotFoundError, FileIOException):
logger.warning(
f"Couldn't load metadata for policy '{cls.__name__}' as the persisted "
f"metadata couldn't be loaded.")
return cls(
config,
model_storage,
resource,
execution_context,
featurizer=featurizer,
lookup=lookup,
)
def load(
cls,
path: Union[Text, Path],
should_finetune: bool = False,
epoch_override: int = defaults[EPOCHS],
**kwargs: Any,
) -> "TEDPolicy":
"""Loads a policy from the storage.
Args:
path: Path on disk where policy is persisted.
should_finetune: Whether to load the policy for finetuning.
epoch_override: Override the number of epochs in persisted
configuration for further finetuning.
**kwargs: Any other arguments
Returns:
Loaded policy
Raises:
`PolicyModelNotFound` if the model is not found in the supplied `path`.
"""
model_path = Path(path)
if not model_path.exists():
logger.warning(
f"Failed to load {cls.__class__.__name__} model. Path "
f"'{model_path.absolute()}' doesn't exist.")
return cls()
featurizer = TrackerFeaturizer.load(path)
if not (model_path /
f"{cls._metadata_filename()}.data_example.pkl").is_file():
return cls(featurizer=featurizer)
model_utilities = cls._load_model_utilities(model_path)
model_utilities["meta"] = cls._update_loaded_params(
model_utilities["meta"])
if should_finetune:
model_utilities["meta"][EPOCHS] = epoch_override
(
model_data_example,
predict_data_example,
) = cls._construct_model_initialization_data(
model_utilities["loaded_data"])
model = cls._load_tf_model(
model_utilities,
model_data_example,
predict_data_example,
featurizer,
should_finetune,
)
return cls._load_policy_with_model(model, featurizer, model_utilities,
should_finetune)
def load(cls, path: Union[Text, Path]) -> "Policy":
"""Loads a policy from path.
Args:
path: Path to load policy from.
Returns:
An instance of `Policy`.
"""
metadata_file = Path(path) / cls._metadata_filename()
if metadata_file.is_file():
data = json.loads(rasa.shared.utils.io.read_file(metadata_file))
if (Path(path) / FEATURIZER_FILE).is_file():
featurizer = TrackerFeaturizer.load(path)
data["featurizer"] = featurizer
return cls(**data)
logger.info(
f"Couldn't load metadata for policy '{cls.__name__}'. "
f"File '{metadata_file}' doesn't exist."
)
return cls()
def test_persist_and_load_tracker_featurizer(tmp_path: Text,
moodbot_domain: Domain):
state_featurizer = SingleStateFeaturizer()
state_featurizer.prepare_for_training(moodbot_domain, RegexInterpreter())
tracker_featurizer = MaxHistoryTrackerFeaturizer(state_featurizer)
tracker_featurizer.persist(tmp_path)
loaded_tracker_featurizer = TrackerFeaturizer.load(tmp_path)
assert loaded_tracker_featurizer is not None
assert loaded_tracker_featurizer.state_featurizer is not None
def load(cls, path: Text) -> "MemoizationPolicy":
featurizer = TrackerFeaturizer.load(path)
memorized_file = os.path.join(path, "memorized_turns.json")
if os.path.isfile(memorized_file):
data = json.loads(rasa.shared.utils.io.read_file(memorized_file))
return cls(featurizer=featurizer,
priority=data["priority"],
lookup=data["lookup"])
else:
logger.info("Couldn't load memoization for policy. "
"File '{}' doesn't exist. Falling back to empty "
"turn memory.".format(memorized_file))
return cls()
def load(cls, path: Union[Text, Path], **kwargs: Any) -> "Policy":
"""Loads a policy from path.
Args:
path: Path to load policy from.
Returns:
An instance of `Policy`.
"""
metadata_file = Path(path) / cls._metadata_filename()
if metadata_file.is_file():
data = json.loads(rasa.shared.utils.io.read_file(metadata_file))
if (Path(path) / FEATURIZER_FILE).is_file():
featurizer = TrackerFeaturizer.load(path)
data["featurizer"] = featurizer
data.update(kwargs)
constructor_args = rasa.shared.utils.common.arguments_of(cls)
if "kwargs" not in constructor_args:
if set(data.keys()).issubset(set(constructor_args)):
rasa.shared.utils.io.raise_deprecation_warning(
f"`{cls.__name__}.__init__` does not accept `**kwargs` "
f"This is required for contextual information e.g. the flag "
f"`should_finetune`.",
warn_until_version="3.0.0",
)
else:
raise UnsupportedDialogueModelError(
f"`{cls.__name__}.__init__` does not accept `**kwargs`. "
f"Attempting to pass {data} to the policy. "
f"This argument should be added to all policies by "
f"Rasa Open Source 3.0.0."
)
return cls(**data)
logger.info(
f"Couldn't load metadata for policy '{cls.__name__}'. "
f"File '{metadata_file}' doesn't exist."
)
return cls()
def load(cls, path: Text) -> "TEDPolicy":
"""Loads a policy from the storage.
**Needs to load its featurizer**
"""
if not os.path.exists(path):
raise Exception(
f"Failed to load TED policy model. Path "
f"'{os.path.abspath(path)}' doesn't exist."
)
model_path = Path(path)
tf_model_file = model_path / f"{SAVE_MODEL_FILE_NAME}.tf_model"
featurizer = TrackerFeaturizer.load(path)
if not (model_path / f"{SAVE_MODEL_FILE_NAME}.data_example.pkl").is_file():
return cls(featurizer=featurizer)
loaded_data = io_utils.pickle_load(
model_path / f"{SAVE_MODEL_FILE_NAME}.data_example.pkl"
)
label_data = io_utils.pickle_load(
model_path / f"{SAVE_MODEL_FILE_NAME}.label_data.pkl"
)
zero_state_features = io_utils.pickle_load(
model_path / f"{SAVE_MODEL_FILE_NAME}.zero_state_features.pkl"
)
label_data = RasaModelData(data=label_data)
meta = io_utils.pickle_load(model_path / f"{SAVE_MODEL_FILE_NAME}.meta.pkl")
priority = io_utils.json_unpickle(
model_path / f"{SAVE_MODEL_FILE_NAME}.priority.pkl"
)
model_data_example = RasaModelData(
label_key=LABEL_KEY, label_sub_key=LABEL_SUB_KEY, data=loaded_data
)
meta = train_utils.update_similarity_type(meta)
model = TED.load(
str(tf_model_file),
model_data_example,
data_signature=model_data_example.get_signature(),
config=meta,
max_history_tracker_featurizer_used=isinstance(
featurizer, MaxHistoryTrackerFeaturizer
),
label_data=label_data,
)
# build the graph for prediction
predict_data_example = RasaModelData(
label_key=LABEL_KEY,
label_sub_key=LABEL_SUB_KEY,
data={
feature_name: features
for feature_name, features in model_data_example.items()
if feature_name
in STATE_LEVEL_FEATURES + FEATURES_TO_ENCODE + [DIALOGUE]
},
)
model.build_for_predict(predict_data_example)
return cls(
featurizer=featurizer,
priority=priority,
model=model,
zero_state_features=zero_state_features,
**meta,
)
def load(
cls,
path: Union[Text, Path],
should_finetune: bool = False,
epoch_override: int = defaults[EPOCHS],
**kwargs: Any,
) -> "TEDPolicy":
"""Loads a policy from the storage.
**Needs to load its featurizer**
"""
model_path = Path(path)
if not model_path.exists():
raise Exception(f"Failed to load TED policy model. Path "
f"'{model_path.absolute()}' doesn't exist.")
tf_model_file = model_path / f"{SAVE_MODEL_FILE_NAME}.tf_model"
featurizer = TrackerFeaturizer.load(path)
if not (model_path /
f"{SAVE_MODEL_FILE_NAME}.data_example.pkl").is_file():
return cls(featurizer=featurizer)
loaded_data = io_utils.pickle_load(
model_path / f"{SAVE_MODEL_FILE_NAME}.data_example.pkl")
label_data = io_utils.pickle_load(
model_path / f"{SAVE_MODEL_FILE_NAME}.label_data.pkl")
zero_state_features = io_utils.pickle_load(
model_path / f"{SAVE_MODEL_FILE_NAME}.zero_state_features.pkl")
label_data = RasaModelData(data=label_data)
meta = io_utils.pickle_load(model_path /
f"{SAVE_MODEL_FILE_NAME}.meta.pkl")
priority = io_utils.json_unpickle(
model_path / f"{SAVE_MODEL_FILE_NAME}.priority.pkl")
model_data_example = RasaModelData(label_key=LABEL_KEY,
label_sub_key=LABEL_SUB_KEY,
data=loaded_data)
meta = train_utils.update_similarity_type(meta)
meta[EPOCHS] = epoch_override
model = TED.load(
str(tf_model_file),
model_data_example,
data_signature=model_data_example.get_signature(),
config=meta,
max_history_tracker_featurizer_used=isinstance(
featurizer, MaxHistoryTrackerFeaturizer),
label_data=label_data,
finetune_mode=should_finetune,
)
if not should_finetune:
# build the graph for prediction
features_to_select = STATE_LEVEL_FEATURES + FEATURES_TO_ENCODE + [
DIALOGUE
]
predict_data_example = RasaModelData(
label_key=LABEL_KEY,
label_sub_key=LABEL_SUB_KEY,
data={
feature_name: features
for feature_name, features in model_data_example.items()
if feature_name in features_to_select
},
)
model.build_for_predict(predict_data_example)
return cls(
featurizer=featurizer,
priority=priority,
model=model,
zero_state_features=zero_state_features,
should_finetune=should_finetune,
**meta,
)
def test_fail_to_load_non_existent_featurizer():
assert TrackerFeaturizer.load("non_existent_class") is None