Hugging Face Plugin for Streamlined Development

Hugging Face Plugin

Welcome to the Hugging Face Plugin documentation! This plugin is designed to supercharge your development workflow, enabling you to accomplish tasks faster and more efficiently. With a single language model, you can build a wide range of tools and automate repetitive coding tasks.

This is colab notebook link :

Installation

To get started, follow these steps:

Clone the Hugging Face Plugin repository:

git clone https://github.com/CASIA-IVA-Lab/FastSAM.git

Download the pretrained weights:

wget https://huggingface.co/spaces/An-619/FastSAM/resolve/main/weights/FastSAM.pt

Install the required dependencies:

pip install -r FastSAM/requirements.txt
pip install git+https://github.com/openai/CLIP.git

Setup

Before diving into the amazing features of this plugin, ensure that you have set up your environment correctly. Follow these steps:

Set the transformers_version to the desired version (e.g., "v4.29.0") using the provided dropdown menu.

Install the necessary libraries and dependencies:

pip install huggingface_hub>=0.14.1 git+https://github.com/huggingface/transformers@$transformers_version -q diffusers accelerate datasets torch soundfile sentencepiece opencv-python openai

Authenticate with your Hugging Face account:

from huggingface_hub import notebook_login
notebook_login()

Initialize the agent by providing the required credentials:

For StarCoder (HF Token):

from transformers.tools import HfAgent
agent = HfAgent("https://api-inference.huggingface.co/models/bigcode/starcoder", token=token)
print("StarCoder is initialized 💪")

For OpenAssistant (HF Token):

from transformers.tools import HfAgent
agent = HfAgent(url_endpoint="https://api-inference.huggingface.co/models/OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5", token=token)
print("OpenAssistant is initialized 💪")

For OpenAI (API Key):

from transformers.tools import OpenAiAgent
pswd = getpass.getpass('OpenAI API key:')
agent = OpenAiAgent(model="text-davinci-003", api_key=pswd)
print("OpenAI is initialized 💪")

Examples

Customer Information Agent

You can create a customer information agent using a DataFrame. Here's an example:

import pandas as pd

data = [
    {"Name": "John", "Last Name": "Smith", "Working Status": "Employed", "Rating Score": 4},
    {"Name": "Emma", "Last Name": "Johnson", "Working Status": "Employed", "Rating Score": 3},
    {"Name": "Alex", "Last Name": "Brown", "Working Status": "Unemployed", "Rating Score": 2},
    {"Name": "Olivia", "Last Name": "Davis", "Working Status": "Employed", "Rating Score": 5},
    {"Name": "Olivia", "Last Name": "Brown", "Working Status": "Employed", "Rating Score": 6},
    {"Name": "Liam", "Last Name": "Wilson", "Working Status": "Employed", "Rating Score": 1}
]

df = pd.DataFrame(data)

def get_customer_info(name, last_name=None):
    if last_name is None:
        filtered_df = df[df['Name'] == name]
    else:
        filtered_df = df[(df['Name'] == name) & (df['Last Name'] == last_name)]

    if len(filtered_df) == 0:
        return "No customer found."
    else:
        return filtered_df.to_dict('records')

Fast Sam Agent

You can perform image segmentation using the Fast Sam model. Here's an example:

from PIL import Image

class FastSamTool(Tool):
    name = "FastSamTool"
    description = ("Returns image segmentation of an image. It takes 'image' and 'text_prompt' as inputs and returns the segmented image.")

    inputs = ["text", "text"]
    outputs = ["image"]

    def __call__(self, image: str, text_prompt: str):
        command = f"python FastSAM/Inference.py --model_path FastSAM.pt --img_path ./images/{image} --text_prompt '{text_prompt}' --imgsz 1024"
        !{command}

        image = Image.open(f"./output/{image}")
        return image

Conclusion

With the Hugging Face Plugin, you have the power to streamline your development process and automate various tasks using a single language model. Explore the provided examples and unleash your creativity to build your own efficient tools. Happy coding!

Please note that the above examples are just a glimpse of what you can achieve with the Hugging Face Plugin. For more details and additional functionalities, refer to the complete documentation in the GitHub repository.