AnyGen: A Unified Interface for Text Generation

AnyGen is a minimal Python library that unifies text generation tasks using Hugging Face, OpenAI, and Gemini models. It offers a minimalistic and unified pipeline for loading models and generating outputs with ease and efficiency.

Features

• Support for Hugging Face models
• Support for OpenAI's GPT models
• Support for Gemini models
• Easy-to-use interface for text generation

Installation

Using pip

You can install AnyGen from PyPI:

pip install -U anygen

From Source

Clone the repository and install it manually:

git clone https://github.com/macabdul9/AnyGen.git
cd AnyGen
pip install .

Requirements

Ensure the following libraries are installed:

pip install transformers google-generativeai requests openai

Usage

Below are step-by-step instructions to generate text using each model type.

1. Hugging Face Model

from anygen import AnyGen

# Initialize the generator
model_name_or_path = "meta-llama/Llama-3.2-1B-Instruct"  # Replace with your Hugging Face model name
device = "cuda"  # Use "cpu" if GPU is not available
hf_generator = AnyGen(model_type="huggingface", model_name_or_path=model_name_or_path, device=device)

# Generate text
prompt = "Write python code for binary search"
generated_text = hf_generator.generate(prompt)
print(generated_text)

2. OpenAI Model

from anygen import AnyGen

# Initialize the generator
api_key_fp = "openai_keys.json"  # Path to your OpenAI credentials file
openai_generator = AnyGen(model_type="openai", api_key_fp=api_key_fp)

# Generate text
prompt = "Write python code for binary search"
generated_text = openai_generator.generate(prompt)
print(generated_text)

3. Gemini Model

from anygen import AnyGen

# Initialize the generator
api_key_fp = "gemini_keys.json"  # Path to your Gemini credentials file
gemini_generator = AnyGen(model_type="gemini", api_key_fp=api_key_fp)

# Generate text
prompt = "Write python code for binary search"
generated_text = gemini_generator.generate(prompt)
print(generated_text)

Example with Parameters

from anygen import AnyGen

# Initialize the generator
api_key_fp = "openai_keys.json"  # Example for OpenAI
openai_generator = AnyGen(model_type="openai", api_key_fp=api_key_fp)

# Generate text with parameters
prompt = "Write python code for binary search"
parameters = {"temperature": 0.7, "max_tokens": 512}
generated_text = openai_generator.generate(prompt, parameters)
print(generated_text)

API Key File Format

Both OpenAI and Gemini models require an API key stored in a JSON file. Below is an example format:

openai_keys.json:

{
    "openai_model_name": {
        "api_key": "your_openai_api_key",
        "endpoint": "your_endpoint"
    }
}

Replace openai_model_name with the OpenAI model name (e.g., gpt-4o-mini), api_key with your API key, and your_endpoint with the provided endpoint URL.

gemini_keys.json:

{
    "gemini_model_name": {
        "api_key": "your_gemini_api_key"
    }
}

Replace gemini_model_name with the Gemini model name (e.g., gemini-2.0-flash-exp), api_key with your Gemini API key.

Parameters

• temperature: Controls the randomness of the output. Higher values produce more random results.
• max_tokens: The maximum number of tokens to generate.
• top_p: The cumulative probability of the top tokens to sample from.
• top_k: The number of highest probability vocabulary tokens to keep for top-k-filtering.
• beam_size: The number of beams to use for beam search.

Running Tests

You can run the tests using the following commands.

1. Clone the repository:

git clone [email protected]:macabdul9/AnyGen.git
cd AnyGen

2. Install the requirements:

pip install -r requirements.txt

3. Run the tests:

python -m tests.test_anygen

Contributions

Feel free to submit issues and/or contribute to this repository!

License

This project is licensed under the MIT License.

Cite this Work

If you use AnyGen in your research or work, please cite it using the following BibTeX entry:

@software{anygen,
  author = {Abdul Waheed},
  title = {AnyGen: A Unified Interface for Text Generation},
  year = {2024},
  publisher = {GitHub},
  url = {https://github.com/macabdul9/AnyGen},
  doi = {https://doi.org/10.5281/zenodo.14533072}
}