finetune_mal_ben.py

#https://huggingface.co/docs/transformers/v4.17.0/en/tasks/sequence_classification
from transformers import Trainer, TrainingArguments
from transformers import AutoTokenizer
from transformers import AutoModelForSequenceClassification,BertForSequenceClassification
from datasets import load_dataset
import numpy as np
import evaluate
from huggingface_hub import HfFolder
tokenizer = AutoTokenizer.from_pretrained("roberta-large")
file_dict = {
  "train" : "benmal.csv",
  "test" :"benmal.csv"
 
}

dataset=load_dataset(
  'csv',
  data_files=file_dict,
  delimiter=',',
  column_names=['text', 'label'],
  skiprows=1
)
raw_dataset=dataset.shuffle()	
def tokenize(batch):
    return tokenizer(batch['text'], padding='max_length', truncation=True, return_tensors="pt")
tokenized_dataset = raw_dataset.map(tokenize, batched=True,remove_columns=["text"])

model_id = "roberta-large"

model =  AutoModelForSequenceClassification.from_pretrained(
     model_id, num_labels=2, ignore_mismatched_sizes=True
)
metric = evaluate.load("f1")

def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    predictions = np.argmax(predictions, axis=1)
    return metric.compute(predictions=predictions, references=labels, average="weighted")

from transformers import DataCollatorWithPadding

data_collator = DataCollatorWithPadding(tokenizer=tokenizer)
repository_id = "datasetsANDmodels/benginVSmaliuos"
    
training_args= TrainingArguments(
    output_dir=repository_id,
    per_device_train_batch_size=8,
    per_device_eval_batch_size=8,
    learning_rate=2e-5,
    num_train_epochs=10,
#	torch_compile=True,
    evaluation_strategy="epoch",
    save_strategy="epoch",
    save_total_limit=2,
    load_best_model_at_end=True,
 #   metric_for_best_model="f1",
#    report_to="tensorboard",
    push_to_hub=True,
    hub_strategy="every_save",
    hub_model_id=repository_id,
    hub_token=HfFolder.get_token(),

)

trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=tokenized_dataset["train"],
    eval_dataset=tokenized_dataset["train"],
 #   compute_metrics=compute_metrics,
#	tokenizer=tokenizer,
#   data_collator=data_collator,
)
import torch._dynamo
torch._dynamo.config.suppress_errors = True
trainer.train()
tokenizer.save_pretrained(repository_id)
trainer.create_model_card()
trainer.push_to_hub()