def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config)
self.classifier_t1 = RobertaClassificationHead(config)
self.classifier_t2 = RobertaClassificationHead(config)
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config, add_pooling_layer=False)
self.classifier = RobertaClassificationHead(config)
self.specific_classifier = RobertaClassificationHead(config)
self.init_weights()
def __init__(self, config):
super().__init__(config)
# self.num_labels_t1 = config.num_labels_t1
# self.num_labels_t2 = config.num_labels_t2
self.num_labels = config.num_labels
self.num_labels_t1 = self.num_labels_t2 = self.num_labels
self.roberta = RobertaModel(config)
# config.num_labels = deepcopy(config.num_labels_t1)
self.classifier_t1 = RobertaClassificationHead(config)
# config.num_labels = deepcopy(config.num_labels_t2)
self.classifier_t2 = RobertaClassificationHead(config)
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config)
self.tfidf = None
self.max_length = config.max_length
self.classifier = RobertaClassificationHead(config)
def __init__(self, config):
super(RobertaForMultipleChoice, self).__init__(config)
self.roberta = RobertaModel(config)
self.classifier = RobertaClassificationHead(config)
self.init_weights
def __init__(self, config):
super(RobertaForXMLC, self).__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config)
self.classifier = RobertaClassificationHead(config)
self.loss_fct = HingeLoss(margin=1.0, squared=True)
def __init__(self, config, pos_weight=None):
super(RobertaForMultiLabelSequenceClassification, self).__init__(config)
self.num_labels = config.num_labels
self.pos_weight = pos_weight
self.roberta = RobertaModel(config)
self.classifier = RobertaClassificationHead(config)
def __init__(self, config, weight=None):
super(CamembertForSequenceClassification, self).__init__(config)
self.num_labels = config.num_labels
self.camembert = CamembertModel(config)
self.classifier = RobertaClassificationHead(config)
self.weight = weight
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.trelm_roberta = TrelmRobertaModel(config)
self.classifier = RobertaClassificationHead(config)
self.init_weights()
def __init__(self, config, weight=None, sliding_window=False):
super(RobertaForSequenceClassification, self).__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config)
self.classifier = RobertaClassificationHead(config)
self.weight = weight
self.sliding_window = sliding_window
def __init__(self, config):
super(ATM, self).__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModelATM(config)
# self.dropout = nn.Dropout(config.hidden_dropout_prob)
# self.classifier = nn.Linear(config.hidden_size, self.num_labels)
self.classifier = RobertaClassificationHead(config)
def __init__(self):
super().__init__()
config = RobertaConfig.from_pretrained('roberta-large',
output_hidden_states=True)
self.roberta = RobertaModel(config=config)
config.num_labels = 1
self.classifier = RobertaClassificationHead(config=config)
# self._debug = 1
self._debug = -1
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config, add_pooling_layer=False)
self.classifier_copa = RobertaClassificationHead(config)
self.classifier_anli = RobertaClassificationHead(config)
self.classifier_snli = RobertaClassificationHead(config)
self.classifier_atomic = RobertaClassificationHead(config)
self.classifier_social = RobertaClassificationHead(config)
self.classifier_hella = RobertaClassificationHead(config)
self.classifier_joci = RobertaClassificationHead(config)
self.init_weights()
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config)
self.classifier = RobertaClassificationHead(config)
self.lm_head = RobertaLMHead(config)
self.init_weights()
# These attributes should be assigned once the model is initialized
self.model_args = None
self.data_args = None
self.label_word_list = None
# For regression
self.lb = None
self.ub = None
# For auto label search.
self.return_full_softmax = None
def main():
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--data_dir",
default=None,
type=str,
required=True,
help=
"The input data dir. Should contain the .tsv files (or other data files) for the task.",
)
parser.add_argument(
"--model_type",
default=None,
type=str,
required=True,
help="Model type selected in the list: " +
", ".join(MODEL_CLASSES.keys()),
)
parser.add_argument(
"--model_name_or_path",
default=None,
type=str,
required=True,
help="Path to pre-trained model or shortcut name selected in the list: "
+ ", ".join(ALL_MODELS),
)
parser.add_argument(
"--task_name",
default=None,
type=str,
required=True,
help="The name of the task to train selected in the list: " +
", ".join(processors.keys()),
)
parser.add_argument(
"--output_dir",
default=None,
type=str,
required=True,
help=
"The output directory where the model predictions and checkpoints will be written.",
)
# Other parameters
parser.add_argument(
"--config_name",
default="",
type=str,
help="Pretrained config name or path if not the same as model_name",
)
parser.add_argument(
"--tokenizer_name",
default="",
type=str,
help="Pretrained tokenizer name or path if not the same as model_name",
)
parser.add_argument(
"--cache_dir",
default="",
type=str,
help=
"Where do you want to store the pre-trained models downloaded from s3",
)
parser.add_argument(
"--max_seq_length",
default=128,
type=int,
help=
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded.",
)
parser.add_argument("--do_train",
action="store_true",
help="Whether to run training.")
parser.add_argument("--do_eval",
action="store_true",
help="Whether to run eval on the dev set.")
parser.add_argument(
"--evaluate_during_training",
action="store_true",
help="Run evaluation during training at each logging step.",
)
parser.add_argument(
"--do_lower_case",
action="store_true",
help="Set this flag if you are using an uncased model.",
)
parser.add_argument(
"--per_gpu_train_batch_size",
default=8,
type=int,
help="Batch size per GPU/CPU for training.",
)
parser.add_argument(
"--per_gpu_eval_batch_size",
default=8,
type=int,
help="Batch size per GPU/CPU for evaluation.",
)
parser.add_argument(
"--gradient_accumulation_steps",
type=int,
default=1,
help=
"Number of updates steps to accumulate before performing a backward/update pass.",
)
parser.add_argument("--learning_rate",
default=5e-5,
type=float,
help="The initial learning rate for Adam.")
parser.add_argument("--weight_decay",
default=0.0,
type=float,
help="Weight decay if we apply some.")
parser.add_argument("--adam_epsilon",
default=1e-8,
type=float,
help="Epsilon for Adam optimizer.")
parser.add_argument("--max_grad_norm",
default=1.0,
type=float,
help="Max gradient norm.")
parser.add_argument(
"--num_train_epochs",
default=3.0,
type=float,
help="Total number of training epochs to perform.",
)
parser.add_argument(
"--max_steps",
default=-1,
type=int,
help=
"If > 0: set total number of training steps to perform. Override num_train_epochs.",
)
parser.add_argument("--warmup_steps",
default=0,
type=int,
help="Linear warmup over warmup_steps.")
parser.add_argument("--logging_steps",
type=int,
default=500,
help="Log every X updates steps.")
parser.add_argument("--save_steps",
type=int,
default=500,
help="Save checkpoint every X updates steps.")
parser.add_argument(
"--eval_all_checkpoints",
action="store_true",
help=
"Evaluate all checkpoints starting with the same prefix as model_name ending and ending with step number",
)
parser.add_argument("--no_cuda",
action="store_true",
help="Avoid using CUDA when available")
parser.add_argument(
"--overwrite_output_dir",
action="store_true",
help="Overwrite the content of the output directory",
)
parser.add_argument(
"--overwrite_cache",
action="store_true",
help="Overwrite the cached training and evaluation sets",
)
parser.add_argument("--seed",
type=int,
default=42,
help="random seed for initialization")
parser.add_argument(
"--fp16",
action="store_true",
help=
"Whether to use 16-bit (mixed) precision (through NVIDIA apex) instead of 32-bit",
)
parser.add_argument(
"--fp16_opt_level",
type=str,
default="O1",
help=
"For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html",
)
parser.add_argument("--local_rank",
type=int,
default=-1,
help="For distributed training: local_rank")
parser.add_argument("--server_ip",
type=str,
default="",
help="For distant debugging.")
parser.add_argument("--server_port",
type=str,
default="",
help="For distant debugging.")
parser.add_argument(
"--calc_final_alignments",
action="store_true",
help="Set this flag if you want to calculate final aligments.",
)
parser.add_argument(
"--calc_alignment_sim_mat",
action="store_true",
help="Set this flag if you want to calculate alignment_sim_mat.",
)
args = parser.parse_args()
if (os.path.exists(args.output_dir) and os.listdir(args.output_dir)
and args.do_train and not args.overwrite_output_dir):
raise ValueError(
"Output directory ({}) already exists and is not empty. Use --overwrite_output_dir to overcome."
.format(args.output_dir))
# Setup distant debugging if needed
if args.server_ip and args.server_port:
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
import ptvsd
print("Waiting for debugger attach")
ptvsd.enable_attach(address=(args.server_ip, args.server_port),
redirect_output=True)
ptvsd.wait_for_attach()
# Setup CUDA, GPU & distributed training
if args.local_rank == -1 or args.no_cuda:
device = torch.device("cuda" if torch.cuda.is_available()
and not args.no_cuda else "cpu")
args.n_gpu = torch.cuda.device_count()
else: # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.cuda.set_device(args.local_rank)
device = torch.device("cuda", args.local_rank)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
level=logging.INFO if args.local_rank in [-1, 0] else logging.WARN,
)
logger.warning(
"Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
args.local_rank,
device,
args.n_gpu,
bool(args.local_rank != -1),
args.fp16,
)
# Set seed
set_seed(args)
# Prepare GLUE task
args.task_name = args.task_name.lower()
if args.task_name not in processors:
raise ValueError("Task not found: %s" % (args.task_name))
processor = processors[args.task_name]()
args.output_mode = output_modes[args.task_name]
label_list = processor.get_labels()
num_labels = len(label_list)
# Load pretrained model and tokenizer
if args.local_rank not in [-1, 0]:
torch.distributed.barrier(
) # Make sure only the first process in distributed training will download model & vocab
args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
if args.model_name_or_path != 'roberta-large-mnli':
config = config_class.from_pretrained(
args.config_name if args.config_name else args.model_name_or_path,
num_labels=num_labels,
finetuning_task=args.task_name,
cache_dir=args.cache_dir if args.cache_dir else None,
)
tokenizer = tokenizer_class.from_pretrained(
args.tokenizer_name
if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None,
)
model = model_class.from_pretrained(
args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None,
)
# for num_labels(mnli)
if args.model_name_or_path == 'roberta-large-mnli':
num_labels_old = config_class.from_pretrained(
args.model_name_or_path)._num_labels
config = config_class.from_pretrained(
args.config_name if args.config_name else args.model_name_or_path,
num_labels=num_labels_old,
finetuning_task=args.task_name,
cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(
args.tokenizer_name
if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
if num_labels != num_labels_old:
config.num_labels = num_labels_old
model = model_class.from_pretrained(
args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
config.num_labels = num_labels
logger.info('Reintializing model classifier layer...')
model.num_labels = num_labels
model.classifier = RobertaClassificationHead(config)
else:
model = model_class.from_pretrained(
args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
# # num_added_toks = tokenizer.add_tokens(['[START]', '[END]'])
# num_added_toks = tokenizer.add_special_tokens({'additional_special_tokens':['[START]', '[END]']})
# # num_added_toks = tokenizer.add_special_tokens({'additional_special_tokens': ['[CONTEXT]']})
# print('We have added', num_added_toks, 'tokens')
# model.resize_token_embeddings(len(tokenizer))
if args.local_rank == 0:
torch.distributed.barrier(
) # Make sure only the first process in distributed training will download model & vocab
model.to(args.device)
logger.info("Training/evaluation parameters %s", args)
# Training
if args.do_train:
train_dataset = load_and_cache_examples(args,
args.task_name,
tokenizer,
evaluate=False)
global_step, tr_loss = train(args, train_dataset, model, tokenizer)
logger.info(" global_step = %s, average loss = %s", global_step,
tr_loss)
# Saving best-practices: if you use defaults names for the model, you can reload it using from_pretrained()
if args.do_train and (args.local_rank == -1
or torch.distributed.get_rank() == 0):
# Create output directory if needed
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
logger.info("Saving model checkpoint to %s", args.output_dir)
# Save a trained model, configuration and tokenizer using `save_pretrained()`.
# They can then be reloaded using `from_pretrained()`
model_to_save = (model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(args.output_dir)
tokenizer.save_pretrained(args.output_dir)
# Good practice: save your training arguments together with the trained model
torch.save(args, os.path.join(args.output_dir, "training_args.bin"))
# Load a trained model and vocabulary that you have fine-tuned
model = model_class.from_pretrained(args.output_dir)
tokenizer = tokenizer_class.from_pretrained(args.output_dir)
model.to(args.device)
# Evaluation
results = {}
if args.do_eval and args.local_rank in [-1, 0]:
tokenizer = tokenizer_class.from_pretrained(
args.output_dir, do_lower_case=args.do_lower_case)
checkpoints = [args.output_dir]
if args.eval_all_checkpoints:
checkpoints = list(
os.path.dirname(c) for c in sorted(
glob.glob(args.output_dir + "/**/" + WEIGHTS_NAME,
recursive=True)))
logging.getLogger("transformers.modeling_utils").setLevel(
logging.WARN) # Reduce logging
logger.info("Evaluate the following checkpoints: %s", checkpoints)
for checkpoint in checkpoints:
global_step = checkpoint.split(
"-")[-1] if len(checkpoints) > 1 else ""
prefix = checkpoint.split(
"/")[-1] if checkpoint.find("checkpoint") != -1 else ""
model = model_class.from_pretrained(checkpoint)
model.to(args.device)
result = evaluate(args,
model,
tokenizer,
prefix=prefix,
finalEval=True)
result = dict(
(k + "_{}".format(global_step), v) for k, v in result.items())
results.update(result)
return results
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.roberta = LongformerModel(config)
self.classifier = RobertaClassificationHead(config)
def main():
parser = get_parser()
args = parser.parse_args()
if (os.path.exists(args.output_dir) and os.listdir(args.output_dir)
and args.do_train and not args.overwrite_output_dir):
raise ValueError(
"Output directory ({}) already exists and is not empty. Use --overwrite_output_dir to overcome."
.format(args.output_dir))
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
# Setup distant debugging if needed
if args.server_ip and args.server_port:
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
import ptvsd
print("Waiting for debugger attach")
ptvsd.enable_attach(address=(args.server_ip, args.server_port),
redirect_output=True)
ptvsd.wait_for_attach()
# Setup CUDA, GPU & distributed training
if args.local_rank == -1 or args.no_cuda:
device = torch.device("cuda" if torch.cuda.is_available()
and not args.no_cuda else "cpu")
args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
else: # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.cuda.set_device(args.local_rank)
device = torch.device("cuda", args.local_rank)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
level=logging.INFO if args.local_rank in [-1, 0] else logging.WARN,
)
logger.warning(
"Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
args.local_rank,
device,
args.n_gpu,
bool(args.local_rank != -1),
args.fp16,
)
# Set seed
set_seed(args.seed)
args.task_name = args.task_name.lower()
if args.task_name not in processors:
raise ValueError("Task not found: %s" % (args.task_name))
processor = processors[args.task_name]()
args.output_mode = output_modes[args.task_name]
label_list = processor.get_labels()
num_labels = len(label_list)
# Load pretrained model and tokenizer
if args.local_rank not in [-1, 0]:
torch.distributed.barrier(
) # Make sure only the first process in distributed training will download model & vocab
args.model_type = args.model_type.lower()
num_labels_old = AutoConfig.from_pretrained(
args.model_name_or_path).num_labels
config = AutoConfig.from_pretrained(
args.config_name if args.config_name else args.model_name_or_path,
num_labels=num_labels_old,
finetuning_task=args.task_name,
cache_dir=args.cache_dir if args.cache_dir else None,
)
tokenizer = AutoTokenizer.from_pretrained(
args.tokenizer_name
if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None,
)
if args.model_type == "electra":
model = ElectraForSequenceClassification.from_pretrained(
args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None,
)
else:
model = AutoModelForSequenceClassification.from_pretrained(
args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None,
)
if num_labels != num_labels_old:
config.num_labels = num_labels
model.num_labels = num_labels
if args.model_type in ["roberta", "bert", "electra"]:
from transformers.modeling_roberta import RobertaClassificationHead
model.classifier = (RobertaClassificationHead(config)
if args.model_type == "roberta" else nn.Linear(
config.hidden_size, config.num_labels))
for module in model.classifier.modules():
if isinstance(module, (nn.Linear, nn.Embedding)):
# Slightly different from the TF version which uses truncated_normal for initialization
# cf https://github.com/pytorch/pytorch/pull/5617
module.weight.data.normal_(mean=0.0,
std=config.initializer_range)
if isinstance(module, nn.Linear) and module.bias is not None:
module.bias.data.zero_()
elif args.model_type == "bart":
from transformers.modeling_bart import BartClassificationHead
model.classification_head = BartClassificationHead(
config.d_model,
config.d_model,
config.num_labels,
config.classif_dropout,
)
model.model._init_weights(model.classification_head.dense)
model.model._init_weights(model.classification_head.out_proj)
elif args.model_type == "xlnet":
model.logits_proj = nn.Linear(config.d_model, config.num_labels)
model.transformer._init_weights(model.logits_proj)
else:
raise NotImplementedError
if args.local_rank == 0:
torch.distributed.barrier(
) # Make sure only the first process in distributed training will download model & vocab
if args.reinit_pooler:
if args.model_type in ["bert", "roberta"]:
encoder_temp = getattr(model, args.model_type)
encoder_temp.pooler.dense.weight.data.normal_(
mean=0.0, std=encoder_temp.config.initializer_range)
encoder_temp.pooler.dense.bias.data.zero_()
for p in encoder_temp.pooler.parameters():
p.requires_grad = True
elif args.model_type in ["xlnet", "bart", "electra"]:
raise ValueError(
f"{args.model_type} does not have a pooler at the end")
else:
raise NotImplementedError
if args.reinit_layers > 0:
if args.model_type in ["bert", "roberta", "electra"]:
assert args.reinit_pooler or args.model_type == "electra"
from transformers.modeling_bert import BertLayerNorm
encoder_temp = getattr(model, args.model_type)
for layer in encoder_temp.encoder.layer[-args.reinit_layers:]:
for module in layer.modules():
if isinstance(module, (nn.Linear, nn.Embedding)):
# Slightly different from the TF version which uses truncated_normal for initialization
# cf https://github.com/pytorch/pytorch/pull/5617
module.weight.data.normal_(
mean=0.0,
std=encoder_temp.config.initializer_range)
elif isinstance(module, BertLayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
if isinstance(module,
nn.Linear) and module.bias is not None:
module.bias.data.zero_()
elif args.model_type == "xlnet":
from transformers.modeling_xlnet import XLNetLayerNorm, XLNetRelativeAttention
for layer in model.transformer.layer[-args.reinit_layers:]:
for module in layer.modules():
if isinstance(module, (nn.Linear, nn.Embedding)):
# Slightly different from the TF version which uses truncated_normal for initialization
# cf https://github.com/pytorch/pytorch/pull/5617
module.weight.data.normal_(
mean=0.0,
std=model.transformer.config.initializer_range)
if isinstance(module,
nn.Linear) and module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, XLNetLayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
elif isinstance(module, XLNetRelativeAttention):
for param in [
module.q,
module.k,
module.v,
module.o,
module.r,
module.r_r_bias,
module.r_s_bias,
module.r_w_bias,
module.seg_embed,
]:
param.data.normal_(
mean=0.0,
std=model.transformer.config.initializer_range)
elif args.model_type == "bart":
for layer in model.model.decoder.layers[-args.reinit_layers:]:
for module in layer.modules():
model.model._init_weights(module)
else:
raise NotImplementedError
if args.mixout > 0:
from mixout import MixLinear
for sup_module in model.modules():
for name, module in sup_module.named_children():
if isinstance(module, nn.Dropout):
module.p = 0.0
if isinstance(module, nn.Linear):
target_state_dict = module.state_dict()
bias = True if module.bias is not None else False
new_module = MixLinear(module.in_features,
module.out_features, bias,
target_state_dict["weight"],
args.mixout)
new_module.load_state_dict(target_state_dict)
setattr(sup_module, name, new_module)
print(model)
model.to(args.device)
logger.info("Training/evaluation parameters %s", args)
# Training
if args.do_train:
train_dataset = load_and_cache_examples(args,
args.task_name,
tokenizer,
evaluate=False)
global_step, tr_loss = train(args, train_dataset, model, tokenizer)
logger.info(" global_step = %s, average loss = %s", global_step,
tr_loss)
# Saving best-practices: if you use defaults names for the model, you can reload it using from_pretrained()
if args.do_train and (args.local_rank == -1
or torch.distributed.get_rank() == 0):
logger.info("Saving model checkpoint to %s", args.output_dir)
torch.save(args, os.path.join(args.output_dir, "training_args.bin"))
def __init__(self,
vocab: Vocabulary,
pretrained_model: str = None,
requires_grad: bool = True,
transformer_weights_model: str = None,
reset_classifier: bool = False,
binary_loss: bool = False,
layer_freeze_regexes: List[str] = None,
on_load: bool = False,
regularizer: Optional[RegularizerApplicator] = None) -> None:
super().__init__(vocab, regularizer)
if on_load:
logging.info(f"Skipping loading of initial Transformer weights")
transformer_config = RobertaConfig.from_pretrained(
pretrained_model)
self._transformer_model = RobertaModel(transformer_config)
elif transformer_weights_model:
logging.info(
f"Loading Transformer weights model from {transformer_weights_model}"
)
transformer_model_loaded = load_archive(transformer_weights_model)
self._transformer_model = transformer_model_loaded.model._transformer_model
else:
self._transformer_model = RobertaModel.from_pretrained(
pretrained_model)
for name, param in self._transformer_model.named_parameters():
grad = requires_grad
if layer_freeze_regexes and grad:
grad = not any(
[bool(re.search(r, name)) for r in layer_freeze_regexes])
param.requires_grad = grad
transformer_config = self._transformer_model.config
self._output_dim = transformer_config.hidden_size
classifier_input_dim = self._output_dim
classifier_output_dim = 1
transformer_config.num_labels = classifier_output_dim
self._classifier = None
if not on_load and transformer_weights_model \
and hasattr(transformer_model_loaded.model, "_classifier") \
and not reset_classifier:
self._classifier = transformer_model_loaded.model._classifier
old_dims = (self._classifier.dense.in_features,
self._classifier.out_proj.out_features)
new_dims = (classifier_input_dim, classifier_output_dim)
if old_dims != new_dims:
logging.info(
f"NOT copying Transformer classifier weights, incompatible dims: {old_dims} vs {new_dims}"
)
self._classifier = None
if self._classifier is None:
self._classifier = RobertaClassificationHead(transformer_config)
self._binary_loss = binary_loss
self._accuracy = CategoricalAccuracy()
self._sigmoid = torch.nn.Sigmoid()
if self._binary_loss:
self._loss = torch.nn.BCEWithLogitsLoss()
else:
self._loss = torch.nn.CrossEntropyLoss()
self._debug = 2
self._padding_value = 1 # The index of the RoBERTa padding token
def __init__(self, config):
super(RobertaForXMLC, self).__init__(config)
self.num_labels = config.num_labels
self.roberta = RobertaModel(config)
self.classifier = RobertaClassificationHead(config)