def get_parser(desc, default_task="translation"):
# Before creating the true parser, we need to import optional user module
# in order to eagerly import custom tasks, optimizers, architectures, etc.
usr_parser = argparse.ArgumentParser(add_help=False, allow_abbrev=False)
usr_parser.add_argument("--user-dir", default=None)
usr_args, _ = usr_parser.parse_known_args()
utils.import_user_module(usr_args)
parser = argparse.ArgumentParser(allow_abbrev=False)
gen_parser_from_dataclass(parser, CommonConfig())
from fairseq.registry import REGISTRIES
for registry_name, REGISTRY in REGISTRIES.items():
parser.add_argument(
"--" + registry_name.replace("_", "-"),
default=REGISTRY["default"],
choices=REGISTRY["registry"].keys(),
)
# Task definitions can be found under fairseq/tasks/
from fairseq.tasks import TASK_REGISTRY
parser.add_argument(
"--task",
metavar="TASK",
default=default_task,
choices=TASK_REGISTRY.keys(),
help="task",
)
# fmt: on
return parser
class InferConfig(FairseqDataclass):
task: Any = None
decoding: DecodingConfig = DecodingConfig()
common: CommonConfig = CommonConfig()
common_eval: CommonEvalConfig = CommonEvalConfig()
checkpoint: CheckpointConfig = CheckpointConfig()
generation: GenerationConfig = GenerationConfig()
distributed_training: DistributedTrainingConfig = DistributedTrainingConfig(
)
dataset: DatasetConfig = DatasetConfig()
class InferConfig(FairseqDataclass):
task: Any = None
decoding: DecodingConfig = DecodingConfig()
common: CommonConfig = CommonConfig()
common_eval: CommonEvalConfig = CommonEvalConfig()
checkpoint: CheckpointConfig = CheckpointConfig()
distributed_training: DistributedTrainingConfig = DistributedTrainingConfig()
dataset: DatasetConfig = DatasetConfig()
is_ax: bool = field(
default=False,
metadata={
"help": "if true, assumes we are using ax for tuning and returns a tuple for ax to consume"
},
)
