Fine tuning transformer models for domain text classification

                  from datasets import Dataset
from transformers import AutoTokenizer, AutoModelForSequenceClassification, TrainingArguments, Trainer

model_name = 'distilbert-base-uncased'
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=3)

def tokenize(batch):
    return tokenizer(batch['text'], truncation=True, padding='max_length', max_length=256)

train_dataset = Dataset.from_pandas(train_df[['text', 'label']]).map(tokenize, batched=True)
valid_dataset = Dataset.from_pandas(valid_df[['text', 'label']]).map(tokenize, batched=True)

args = TrainingArguments(
    output_dir='artifacts/topic_classifier',
    evaluation_strategy='epoch',
    save_strategy='epoch',
    learning_rate=2e-5,
    per_device_train_batch_size=16,
    per_device_eval_batch_size=32,
    num_train_epochs=3,
    load_best_model_at_end=True,
)

trainer = Trainer(model=model, args=args, train_dataset=train_dataset, eval_dataset=valid_dataset)
trainer.train()