title_gen.py

import random
import numpy as np
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import spaces

# Available models
MODEL_OPTIONS_TITLE = {
    "V6 Model": "alakxender/t5-divehi-title-generation-v6",
    "XS Model": "alakxender/t5-dhivehi-title-generation-xs"
}

# Cache for loaded models/tokenizers
MODEL_CACHE = {}

def get_model_and_tokenizer(model_dir):
    if model_dir not in MODEL_CACHE:
        print(f"Loading model: {model_dir}")
        tokenizer = AutoTokenizer.from_pretrained(model_dir)
        model = AutoModelForSeq2SeqLM.from_pretrained(model_dir)
        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        print(f"Moving model to device: {device}")
        model.to(device)
        MODEL_CACHE[model_dir] = (tokenizer, model)
    return MODEL_CACHE[model_dir]

prefix = "2title: "

max_input_length = 512
max_target_length = 32

@spaces.GPU()
def generate_title(content, seed, use_sampling, model_choice):
    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    if torch.cuda.is_available():
        torch.cuda.manual_seed_all(seed)

    model_dir = MODEL_OPTIONS_TITLE[model_choice]
    tokenizer, model = get_model_and_tokenizer(model_dir)

    input_text = prefix + content.strip()
    inputs = tokenizer(
        input_text,
        max_length=max_input_length,
        truncation=True,
        return_tensors="pt"
    )
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    inputs = {k: v.to(device) for k, v in inputs.items()}

    gen_kwargs = {
        "input_ids": inputs["input_ids"],
        "attention_mask": inputs["attention_mask"],
        "max_length": max_target_length,
        "no_repeat_ngram_size": 2,
    }

    if use_sampling:
        gen_kwargs.update({
            "do_sample": True,
            "temperature": 1.0,
            "top_p": 0.95,
            "num_return_sequences": 1,
        })
    else:
        gen_kwargs.update({
            "num_beams": 4,
            "do_sample": False,
            "early_stopping": True,
        })

    with torch.no_grad():
        outputs = model.generate(**gen_kwargs)
    title = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return title