app.py

import os
import urllib.request
import gradio as gr
from transformers import T5Tokenizer, T5ForConditionalGeneration
import huggingface_hub
import re
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
import time
import transformers
import requests
import globals
from utility import *

"""set up"""
huggingface_hub.login(token=globals.HF_TOKEN)
gemma_tokenizer = AutoTokenizer.from_pretrained(globals.gemma_2b_URL)
gemma_model = AutoModelForCausalLM.from_pretrained(globals.gemma_2b_URL)

falcon_tokenizer = AutoTokenizer.from_pretrained(globals.falcon_7b_URL, trust_remote_code=True, device_map=globals.device_map, offload_folder="offload")
falcon_model = AutoModelForCausalLM.from_pretrained(globals.falcon_7b_URL, trust_remote_code=True,
                                                    torch_dtype=torch.bfloat16, device_map=globals.device_map, offload_folder="offload")

def get_model(model_typ):
  if model_typ not in ["gemma", "falcon", "falcon_api", "simplet5_base", "simplet5_large"]:
    raise ValueError('Invalid model type. Choose "gemma", "falcon", "falcon_api","simplet5_base", "simplet5_large".')
  if model_typ=="gemma":
    tokenizer = gemma_tokenizer
    model = gemma_model
    prefix = globals.gemma_PREFIX
  elif model_typ=="falcon_api":
    prefix = globals.falcon_PREFIX
    model=None
    tokenizer = None
  elif model_typ=="falcon":
    tokenizer = falcon_tokenizer
    model = falcon_model
    prefix = globals.falcon_PREFIX
  elif model_typ in ["simplet5_base","simplet5_large"]:
    prefix = globals.simplet5_PREFIX
    URL = globals.simplet5_base_URL if model_typ=="simplet5_base" else globals.simplet5_large_URL
    T5_MODEL_PATH = f"https://huggingface.co/{URL}/resolve/main/{globals.T5_FILE_NAME}"
    fetch_model(T5_MODEL_PATH, globals.T5_FILE_NAME)
    tokenizer = T5Tokenizer.from_pretrained(URL)
    model = T5ForConditionalGeneration.from_pretrained(URL)
  return model, tokenizer, prefix

def single_query(model_typ="gemma",prompt="She has a heart of gold",
                 max_length=256,
                 api_token=""):
  model, tokenizer, prefix = get_model(model_typ)
  if api_token=="" and model_typ=="falcon_api":
    return "Warning: Aborted, Access token needed to access HuggingFace FalconAPI"

  start_time = time.time()
  input = prefix.replace("{fig}", prompt)
  print(f"Input to model: \n{input}")

  if model_typ == "simplet5_base" or model_typ == "simplet5_large":
    inputs = tokenizer(input, return_tensors="pt")
    outputs = model.generate(
        inputs["input_ids"],
        temperature=0.7,
        max_length=max_length,
        num_beams=5,
        top_k=10,
        do_sample=True,
        num_return_sequences=1,
        early_stopping=True
    )
    answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
  elif model_typ=="gemma":
    inputs = tokenizer(input, return_tensors="pt")
    generate_ids = model.generate(inputs.input_ids, max_length=max_length)
    output= tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
    print(f"Model original output:{output}\n")
    answer = post_process(output,input)
    # pattern = r"\*\*Literal Meaning:\*\*\s*(.*?)(?:\n\n|$)"
    # match = re.search(pattern, output, re.DOTALL)
    # if match:
    #     answer = match.group(1).strip()
    # else:
    #     answer = output

  elif model_typ=="falcon":
    falcon_pipeline = transformers.pipeline(
        "text-generation",
        model=model,
        tokenizer=tokenizer,
    )
    sequences = falcon_pipeline(
        prompt,
        max_length=max_length,
        do_sample=False,  # processing time too long, disable sampling for deterministic output
        num_return_sequences=1,
        eos_token_id=falcon_tokenizer.eos_token_id,
    )
    for seq in sequences:
        print(f"Result: \n{seq['generated_text']}")
  elif model_typ=="falcon_api":
    API_URL = "https://api-inference.huggingface.co/models/tiiuae/falcon-7b-instruct"
    headers = {"Authorization": f"Bearer {api_token}"}
    payload = {
            "inputs": input,
            "parameters": {
                "temperature": 0.7,
                "max_length": max_length,
                "num_return_sequences": 1
            }
        }
    output = api_query(API_URL=API_URL,headers=headers,payload=payload)
    answer = output[0]["generated_text"]
    answer = post_process(answer,input)

  else:
    raise ValueError('Invalid model type. Choose "gemma", "falcon", "falcon_api","simplet5_base", "simplet5_large".')

  print(f"Time taken: {time.time()-start_time:.2f} seconds")
  print(f"processed model output: {answer}")

  return answer

model_types = ["gemma", "falcon", "falcon_api", "simplet5_base", "simplet5_large"]

single_gradio = gr.Interface(
    fn=single_query,
    inputs=[

        gr.Dropdown(choices=model_types, label="Select Model Type"),
        gr.Textbox(lines=2, placeholder="Enter a sentence...", label="Input Sentence"),
        gr.Slider(minimum=50, maximum=512, step=10, value=256, label="Max Length"),
        gr.Textbox(lines=1, placeholder="Enter your API token", label="HuggingFace Token",value=""),
    ],
    outputs="text",
    theme=gr.themes.Soft(),
    title=globals.TITLE,
    description="Select a model type from the dropdown and input a sentence to get the paraphrased literal meaning",
    examples=globals.EXAMPLE
)

if __name__ == '__main__':
    single_gradio.launch()