Spaces:
Sleeping
Sleeping
File size: 1,793 Bytes
7713b1f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 |
import logging
import os
import random
import sys
from transformers import (
AutoConfig,
AutoTokenizer,
)
from model.utils import get_model, TaskType
from tasks.glue.dataset import GlueDataset
from training.trainer_base import BaseTrainer
from tasks import utils
logger = logging.getLogger(__name__)
def get_trainer(args):
model_args, data_args, training_args, _ = args
tokenizer = AutoTokenizer.from_pretrained(
model_args.model_name_or_path,
use_fast=model_args.use_fast_tokenizer,
revision=model_args.model_revision,
)
tokenizer = utils.add_task_specific_tokens(tokenizer)
dataset = GlueDataset(tokenizer, data_args, training_args)
if not dataset.is_regression:
config = AutoConfig.from_pretrained(
model_args.model_name_or_path,
num_labels=dataset.num_labels,
label2id=dataset.label2id,
id2label=dataset.id2label,
finetuning_task=data_args.dataset_name,
revision=model_args.model_revision,
)
else:
config = AutoConfig.from_pretrained(
model_args.model_name_or_path,
num_labels=dataset.num_labels,
finetuning_task=data_args.dataset_name,
revision=model_args.model_revision,
)
model = get_model(model_args, TaskType.SEQUENCE_CLASSIFICATION, config)
# Initialize our Trainer
trainer = BaseTrainer(
model=model,
args=training_args,
train_dataset=dataset.train_dataset if training_args.do_train else None,
eval_dataset=dataset.eval_dataset if training_args.do_eval else None,
compute_metrics=dataset.compute_metrics,
tokenizer=tokenizer,
data_collator=dataset.data_collator,
)
return trainer, None |