Spaces:
Running
Running
File size: 1,549 Bytes
080eba4 |
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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 |
import gradio as gr
from transformers import T5Tokenizer, T5ForConditionalGeneration
from peft import PeftModel
from deep_translator import GoogleTranslator
import torch
# Load tokenizer and base model
tokenizer = T5Tokenizer.from_pretrained("./")
base_model = T5ForConditionalGeneration.from_pretrained("t5-small")
model = PeftModel.from_pretrained(base_model, "./")
# Use GPU if available
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
def summarize(text, language='en'):
if language != 'en':
text = GoogleTranslator(source='auto', target='en').translate(text)
input_ids = tokenizer("summarize: " + text, return_tensors="pt", truncation=True, max_length=512).input_ids.to(device)
output_ids = model.generate(
input_ids,
max_length=80,
min_length=15,
length_penalty=1.5,
num_beams=8,
no_repeat_ngram_size=3,
early_stopping=True
)
summary = tokenizer.decode(output_ids[0], skip_special_tokens=True)
if language != 'en':
summary = GoogleTranslator(source='en', target=language).translate(summary)
return summary
gr.Interface(
fn=summarize,
inputs=[
gr.Textbox(lines=10, label="Enter Article"),
gr.Dropdown(choices=["en", "hi", "te", "fr", "es", "de"], value="en", label="Output Language")
],
outputs="textbox",
title="T5 Summarizer (LoRA Optimized)",
description="Summarize articles using a fine-tuned T5 model with LoRA. Supports multiple languages."
).launch()
|