def create_predictor(
config: PyTextConfig,
model_file: Optional[str] = None,
db_type: str = CAFFE2_DB_TYPE,
task: Optional[NewTask] = None,
) -> Predictor:
"""
Create a simple prediction API from a training config and an exported caffe2
model file. This model file should be created by calling export on a trained
model snapshot.
"""
workspace_id = str(uuid.uuid4())
workspace.SwitchWorkspace(workspace_id, True)
predict_net = predictor_exporter.prepare_prediction_net(
filename=model_file or config.export_caffe2_path, db_type=db_type
)
new_task = task or NewTask.from_config(config.task)
input_tensorizers = {
name: tensorizer
for name, tensorizer in new_task.data.tensorizers.items()
if tensorizer.is_input
}
return lambda input: _predict(
workspace_id, predict_net, new_task.model, input_tensorizers, input
)
def create_predictor(config: PyTextConfig,
model_file: Optional[str] = None) -> Predictor:
"""
Create a simple prediction API from a training config and an exported caffe2
model file. This model file should be created by calling export on a trained
model snapshot.
"""
workspace_id = str(uuid.uuid4())
workspace.SwitchWorkspace(workspace_id, True)
predict_net = predictor_exporter.prepare_prediction_net(
filename=model_file or config.export_caffe2_path,
db_type=CAFFE2_DB_TYPE)
supportedInputTensorizers = [
FloatListTensorizer,
GazetteerTensorizer,
TokenTensorizer,
]
new_task = NewTask.from_config(config.task)
input_tensorizers = {
name: tensorizer
for name, tensorizer in new_task.data.tensorizers.items() if any(
isinstance(tensorizer, t) for t in supportedInputTensorizers)
}
return lambda input: _predict(workspace_id, predict_net, new_task.model,
input_tensorizers, input)
def train(
self,
training_data: DataLoader,
eval_data: DataLoader,
model: Model,
optimizer: Optimizer,
label_names: List[str],
scheduler: Scheduler = None,
sparsifier: Sparsifier = None,
metric_reporter: MetricReporter = None,
train_config: PyTextConfig = None,
rank: int = 0,
) -> Tuple[torch.nn.Module, Any]:
# temp workaround to minimize changes to TaskTrainer
if not train_config:
train_config = PyTextConfig(
task=NewTask.Config(model=RoBERTa.Config), version=20)
if scheduler:
self.scheduler = scheduler
if sparsifier:
self.sparsifier = sparsifier
state = TrainingState(
model=model,
optimizer=optimizer,
scheduler=self.scheduler,
sparsifier=self.sparsifier,
rank=rank,
)
metric_reporter_config = ClassificationMetricReporter.Config(
output_path="/tmp/test_out.txt",
pep_format=False,
model_select_metric=ComparableClassificationMetric.
ACCURACY, # in json: "accuracy"
target_label=None,
text_column_names=["text"],
additional_column_names=[],
recall_at_precision_thresholds=[0.2, 0.4, 0.6, 0.8, 0.9],
)
metric_reporter = ClassificationMetricReporter.from_config_and_label_names(
config=metric_reporter_config, label_names=label_names)
return self.train_from_state(state, training_data, eval_data,
metric_reporter, train_config)
def create_predictor(
config: PyTextConfig,
model_file: Optional[str] = None,
db_type: str = CAFFE2_DB_TYPE,
task: Optional[NewTask] = None,
cache_size: int = 0,
) -> Predictor:
"""
Create a simple prediction API from a training config and an exported caffe2
model file. This model file should be created by calling export on a trained
model snapshot.
"""
workspace_id = str(uuid.uuid4())
workspace.SwitchWorkspace(workspace_id, True)
predict_net = predictor_exporter.prepare_prediction_net(
filename=model_file
or PathManager.get_local_path(config.export_caffe2_path),
db_type=db_type,
)
new_task = task or NewTask.from_config(config.task)
input_tensorizers = {
name: tensorizer
for name, tensorizer in new_task.data.tensorizers.items()
if tensorizer.is_input
}
def predict_fn(input):
return _predict(workspace_id, predict_net, new_task.model,
input_tensorizers, input)
if cache_size < 0:
return lru_cache(maxsize=None)(predict_fn)
elif cache_size > 0:
return lru_cache(maxsize=cache_size)(predict_fn)
else:
return predict_fn