def create_optimizer(config_path):
msg.info(f"Loading config from: {config_path}")
config = util.load_config(config_path, create_objects=False)
util.fix_random_seed(config["training"]["seed"])
config = util.load_config(config_path, create_objects=True)
training = config["training"]
return training["optimizer"]
def test_issue7055():
"""Test that fill-config doesn't turn sourced components into factories."""
source_cfg = {
"nlp": {"lang": "en", "pipeline": ["tok2vec", "tagger"]},
"components": {
"tok2vec": {"factory": "tok2vec"},
"tagger": {"factory": "tagger"},
},
}
source_nlp = English.from_config(source_cfg)
with make_tempdir() as dir_path:
# We need to create a loadable source pipeline
source_path = dir_path / "test_model"
source_nlp.to_disk(source_path)
base_cfg = {
"nlp": {"lang": "en", "pipeline": ["tok2vec", "tagger", "ner"]},
"components": {
"tok2vec": {"source": str(source_path)},
"tagger": {"source": str(source_path)},
"ner": {"factory": "ner"},
},
}
base_cfg = Config(base_cfg)
base_path = dir_path / "base.cfg"
base_cfg.to_disk(base_path)
output_path = dir_path / "config.cfg"
fill_config(output_path, base_path, silent=True)
filled_cfg = load_config(output_path)
assert filled_cfg["components"]["tok2vec"]["source"] == str(source_path)
assert filled_cfg["components"]["tagger"]["source"] == str(source_path)
assert filled_cfg["components"]["ner"]["factory"] == "ner"
assert "model" in filled_cfg["components"]["ner"]
def ray_train_cli(
# fmt: off
ctx: typer.Context, # This is only used to read additional arguments
config_path: Path = Arg(..., help="Path to config file", exists=True),
code_path: Optional[Path] = Opt(None, "--code", "-c", help="Path to Python file with additional code (registered functions) to be imported"),
output_path: Optional[Path] = Opt(None, "--output", "--output-path", "-o", help="Output directory or remote storage URL for saving trained pipeline"),
num_workers: int = Opt(1, "--n-workers", "-w", help="Number of workers"),
ray_address: Optional[str] = Opt(None, "--address", "-a", help="Address of ray cluster"),
use_gpu: int = Opt(-1, "--gpu-id", "-g", help="GPU ID or -1 for CPU"),
verbose: bool = Opt(False, "--verbose", "-V", "-VV", help="Display more information for debugging purposes"),
# fmt: on
):
"""
Train a spaCy pipeline using Ray for parallel training.
"""
# TODO: wire up output path
logger.setLevel(logging.DEBUG if verbose else logging.ERROR)
setup_gpu(use_gpu)
overrides = parse_config_overrides(ctx.args)
with show_validation_error(config_path):
config = load_config(config_path, overrides=overrides, interpolate=False)
ray_train(
config,
ray_address=ray_address,
num_workers=num_workers,
use_gpu=use_gpu,
code_path=code_path,
)
def test_create_nlp_from_pretraining_config():
"""Test that the default pretraining config validates properly"""
config = Config().from_str(pretrain_config_string)
pretrain_config = load_config(DEFAULT_CONFIG_PRETRAIN_PATH)
filled = config.merge(pretrain_config)
registry.resolve(filled["pretraining"], schema=ConfigSchemaPretrain)
