def nlp(trf_name_or_path):
config = f"""
lang:
name: en
torch: true
optimizer:
class: torch.optim.SGD
pipeline:
{TRANSFORMERS_TOKENIZER}:
trf_name_or_path: {trf_name_or_path}
"""
return create_model(config)
def nlp(trf_name_or_path, labels, lang, device, textcat_type):
config = f"""
lang:
name: {lang}
torch: true
optimizer:
class: torch.optim.SGD
params:
lr: 0.01
pipeline:
{TRANSFORMERS_SEQ_CLASSIFIER if textcat_type == "single" else TRANSFORMERS_MULTILABEL_SEQ_CLASSIFIER}:
trf_name_or_path: {trf_name_or_path}
labels: {labels}
"""
return create_model(config)
def nlp():
name = "albert-base-v2"
config = f"""
lang:
name: en
optimizer:
class: torch.optim.SGD
params:
lr: 0.01
pipeline:
transformers_model:
trf_name_or_path: {name}
transformers_ner:
labels: {ALL_LABELS}
"""
return create_model(config)
def _main(cfg: Config) -> None:
if cfg.user_config is not None:
# Override config by user config.
# This `user_config` have some limitations, and it will be improved
# after the issue https://github.com/facebookresearch/hydra/issues/386 solved
cfg = OmegaConf.merge(
cfg, OmegaConf.load(hydra.utils.to_absolute_path(cfg.user_config)))
cfg = parse(cfg)
if cfg.seed:
set_seed(cfg.seed)
logger.info(cfg.pretty())
train_data, val_data = create_data(cfg.train.data)
nlp = cast(TorchLanguage, create_model(cfg.model))
logger.info("output dir: {}".format(os.getcwd()))
if torch.cuda.is_available():
logger.info("CUDA enabled")
nlp.to(torch.device("cuda"))
savedir = Path.cwd() / "models"
savedir.mkdir(exist_ok=True)
train(cfg.train, nlp, train_data, val_data, savedir)
def _main(cfg):
cfg = parse(cfg)
if cfg.seed:
set_seed(cfg.seed)
org_cwd = hydra.utils.get_original_cwd()
logger.info(cfg.pretty())
nlp = cast(TorchLanguage, create_model(cfg.model))
train_data = list(
srsly.read_jsonl(os.path.join(org_cwd, cfg.train.data.train)))
cfg.train.data.ndata = len(train_data)
val_data = list(srsly.read_jsonl(os.path.join(org_cwd,
cfg.train.data.val)))
logger.info("output dir: {}".format(os.getcwd()))
if torch.cuda.is_available():
logger.info("CUDA enabled")
nlp.to(torch.device("cuda"))
savedir = Path.cwd() / "models"
srsly.write_jsonl(Path.cwd() / f"train-data.jsonl", train_data)
srsly.write_jsonl(Path.cwd() / f"val-data.jsonl", val_data)
savedir.mkdir(exist_ok=True)
train(cfg.train, nlp, train_data, val_data, savedir)
def nlp_trf_model(trf_model_config, device):
_nlp = create_model(trf_model_config)
_nlp.to(device)
return _nlp
def get_model_value(cfg):
nlp = create_model(cfg.model)
pipe = nlp.get_pipe(TRANSFORMERS_NER)
return sum(p.sum().cpu().item() for p in pipe.model.parameters())
def test_freeze_model(trf_testmodel_path, trf_model_config: NLPConfig):
config = omegaconf.OmegaConf.to_container(trf_model_config)
config["pipeline"][TRANSFORMERS_MODEL]["freeze"] = True
nlp = create_model(config)
pipe = nlp.pipeline[-1][1]
assert pipe.cfg["freeze"]
def nlp(config, device):
_nlp = create_model(config)
_nlp.to(device)
return _nlp
def test_add_pipes_parser(yml, pipe):
nlp = create_model(yml)
assert nlp.get_pipe(pipe)