instruct_dv.py

import random
import numpy as np
import torch
from transformers import AutoTokenizer, T5ForConditionalGeneration
import spaces


# Available models
MODEL_OPTIONS_INSTRUCT = {
    "I2 Model": "alakxender/flan-t5-base-alpaca-dv5",
    "I1 Model": "alakxender/flan-t5-base-alpaca-dv"
}

# 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 = T5ForConditionalGeneration.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]

max_input_length = 256
max_output_length = 256

@spaces.GPU()
def generate_response(instruction, input_text, seed, use_sampling, model_choice,max_tokens,num_beams):
    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_INSTRUCT[model_choice]
    tokenizer, model = get_model_and_tokenizer(model_dir)

    combined_input = f"{instruction.strip()} {input_text.strip()}" if input_text else instruction.strip()
    inputs = tokenizer(
        combined_input,
        return_tensors="pt",
        truncation=True,
        max_length=max_input_length
    )
    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_new_tokens":max_tokens,
        "max_length": max_output_length,
        "no_repeat_ngram_size": 3,
        "repetition_penalty": 1.5,
    }

    if use_sampling:
        gen_kwargs.update({
            "do_sample": True,
            "temperature": 0.1,
            "num_return_sequences": 1,
            "num_beams": 1,
        })
    else:
        gen_kwargs.update({
            "num_beams": 8,
            "do_sample": False,
            "early_stopping": True,
        })

    with torch.no_grad():
        outputs = model.generate(**gen_kwargs)
    decoded_output = tokenizer.decode(outputs[0], skip_special_tokens=True)
    
    # Trim to the last period
    if '.' in decoded_output:
        last_period = decoded_output.rfind('.')
        decoded_output = decoded_output[:last_period+1]

    return decoded_output