File size: 1,108 Bytes
464ed15 |
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 |
import gradio as gr
import torch
from transformers import AutoModelForSequenceClassification, AutoTokenizer, TrainingArguments, Trainer
from datasets import load_dataset
import os
# Function to fine-tune model
def fine_tune(model_name, dataset_url, file, epochs, batch_size, learning_rate):
try:
# Load dataset
if dataset_url:
dataset = load_dataset(dataset_url)
elif file:
dataset = load_dataset("csv", data_files={"train": file.name})
else:
return "Please provide a dataset URL or upload a file."
# Load model & tokenizer
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=2)
tokenizer = AutoTokenizer.from_pretrained(model_name)
def tokenize_function(examples):
return tokenizer(examples["text"], padding="max_length", truncation=True)
dataset = dataset.map(tokenize_function, batched=True)
# Define training arguments
training_args = TrainingArguments(
output_dir="./results",
evaluation_strategy="epoch
|