app.py

import torch
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

peft_model_id = "sksayril/bpt-v-4-Bengali"
model = AutoModelForCausalLM.from_pretrained(peft_model_id, load_in_8bit=False,device_map='auto')
tokenizer = AutoTokenizer.from_pretrained(peft_model_id)

def generate_text(prompt):
    with torch.no_grad():
        input_ids = tokenizer(prompt, return_tensors="pt").input_ids
        output = model.generate(input_ids, max_length=100)
        generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
        return generated_text

iface = gr.Interface(
    fn=generate_text,
    inputs="text",
    outputs="text",
    title="Bengali Text Generation",
    description="Enter a prompt to generate Bengali text.",
)

iface.launch()