app.py

import datetime
import os
import pytz
import requests
import yaml
from huggingface_hub import InferenceClient
from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool
from tools.final_answer import FinalAnswerTool
from tools.visit_webpage import VisitWebpageTool
from tools.web_search import DuckDuckGoSearchTool
from transformers import pipeline, AutoModelForCausalLM
from Gradio_UI import GradioUI
from PIL.PngImagePlugin import PngImageFile

client = InferenceClient(
    provider="auto",
    api_key=os.environ["HF_TOKEN"],
)

@tool
def document_qa(document: str, question: str)-> str:
    """A tool that finds the answer to a question in a specified document.
    Args:
        document: A string representing the URL to the document. The document can be any file type, including PDF, PNG, and JPEG.
        question: A string representing the question being asked.
    """
    pipe = pipeline("document-question-answering", model="impira/layoutlm-document-qa",)
    answer = pipe(document, question)
    return answer[0]['answer']

@tool
def get_current_time_in_timezone(timezone: str) -> str:
    """A tool that fetches the current local time in a specified timezone.
    Args:
        timezone: A string representing a valid timezone (e.g., 'America/New_York').
    """
    try:
        # Create timezone object
        tz = pytz.timezone(timezone)
        # Get current time in that timezone
        local_time = datetime.datetime.now(tz).strftime("%Y-%m-%d %H:%M:%S")
        return f"The current local time in {timezone} is: {local_time}"
    except Exception as e:
        return f"Error fetching time for timezone '{timezone}': {str(e)}"

@tool
def image_generator(prompt: str) -> PngImageFile:
    """A tool that generates an image based on a specified description.
    Args:
        prompt: A string describing the image to generate.
    """
    image = client.text_to_image(
        prompt,
        model="ByteDance/SDXL-Lightning",
    )
    return image

@tool
def image_qa(image: str, question: str) -> str:
    """A tool that answers a question based on a specified image.
    Args:
        image: A string representing the URL to the image.
        question: A string representing the question being asked.
    """
    pipe = pipeline("visual-question-answering", model="dandelin/vilt-b32-finetuned-vqa")
    answer = pipe(image, question)
    return answer[0]['answer']

@tool
def translator(text: str, src_lang: str, tgt_lang: str) -> str:
    """A tool that translates text from a specified language to another specified language.
    Args:
        text: A string to translate.
        src_lang: A string representation of the source language code (e.g., 'fr'). Must be two characters long and in lowercase.
        tgt_lang: A string representation of the target language code (e.g., 'en'). Must be two characters long and in lowercase.
    """
    model = "Helsinki-NLP/opus-mt-" + src_lang + "-" + tgt_lang
    result = client.translation(
        text,
        model=model,
    )
    return result.translation_text

final_answer = FinalAnswerTool()
visit_webpage = VisitWebpageTool()
web_search = DuckDuckGoSearchTool()

# If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
# model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud' 

model = HfApiModel(
    max_tokens=2096,
    temperature=0.5,
    model_id='Qwen/Qwen2.5-Coder-32B-Instruct',# it is possible that this model may be overloaded
    custom_role_conversions=None,
)

with open("prompts.yaml", 'r') as stream:
    prompt_templates = yaml.safe_load(stream)
    
agent = CodeAgent(
    model=model,
    tools=[document_qa, final_answer, image_generator, image_qa, translator, visit_webpage, web_search],
    max_steps=6,
    verbosity_level=1,
    grammar=None,
    planning_interval=None,
    name=None,
    description=None,
    prompt_templates=prompt_templates
)

GradioUI(agent).launch()