app.py

from transformers import XLMRobertaTokenizer, XLMRobertaForSequenceClassification, Trainer, TrainingArguments
import datasets

# Load the pre-trained XLM-Roberta-Large model and tokenizer
model_name = 'xlm-roberta-large'
tokenizer = XLMRobertaTokenizer.from_pretrained(model_name)
model = XLMRobertaForSequenceClassification.from_pretrained(model_name, num_labels=2)

# Load the sentiment analysis dataset
dataset = datasets.load_dataset('imdb')

# Tokenize the dataset
def tokenize(batch):
    return tokenizer(batch['text'], padding=True, truncation=True)

dataset = dataset.map(tokenize, batched=True)

# Fine-tune the model on the dataset
training_args = TrainingArguments(
    output_dir='./results',
    evaluation_strategy='epoch',
    learning_rate=2e-5,
    per_device_train_batch_size=8,
    per_device_eval_batch_size=8,
    num_train_epochs=3,
    weight_decay=0.01,
    push_to_hub=False,
    logging_dir='./logs',
    logging_steps=10,
    load_best_model_at_end=True,
    metric_for_best_model='accuracy'
)

trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=dataset['train'],
    eval_dataset=dataset['test']
)

trainer.train()

import torch

# Load the fine-tuned XLM-Roberta-Large model
model_path = './results/checkpoint-1000'
model = XLMRobertaForSequenceClassification.from_pretrained(model_path)

# Create a function that takes a text input and returns the predicted sentiment label
def predict_sentiment(text):
    inputs = tokenizer(text, padding=True, truncation=True, return_tensors='pt')
    outputs = model(**inputs)
    logits = outputs.logits
    predicted_class = torch.argmax(logits, dim=1)
    return 'positive' if predicted_class == 1 else 'negative'

import gradio as gr

# Create a Gradio interface for the predict_sentiment function
iface = gr.Interface(
    fn=predict_sentiment,
    inputs=gr.inputs.Textbox(placeholder='Enter text here...'),
    outputs=gr.outputs.Textbox(placeholder='Sentiment prediction...')
)

# Launch the interface
iface.launch()