langchain-huggy 0.2.2

A LangChain integration for HuggingChat models

langchain_huggy

langchain_huggy is a Python package that provides an easy-to-use interface for interacting with HuggingChat models through the LangChain framework. It allows you to leverage powerful language models in your applications with minimal setup.

Available Models

langchain_huggy comes with several pre-configured models:

1. 'meta-llama/Meta-Llama-3.1-70B-Instruct'
2. 'CohereForAI/c4ai-command-r-plus-08-2024'
3. 'Qwen/Qwen2.5-72B-Instruct' (default)
4. 'meta-llama/Llama-3.2-11B-Vision-Instruct'
5. 'NousResearch/Hermes-3-Llama-3.1-8B'
6. 'mistralai/Mistral-Nemo-Instruct-2407'
7. 'microsoft/Phi-3.5-mini-instruct'

You can choose any of these models when initializing the HuggingChat instance.

Installation

Install the package using pip:

pip install langchain_huggy

Quick Start

Here's a simple example to get you started:

from langchain_huggy import HuggingChat
from langchain_core.messages import HumanMessage

# Initialize the HuggingChat model
llm = HuggingChat(
    hf_email="[e-mail]",
    hf_password="your_huggingface_password",
    model="Qwen/Qwen2.5-72B-Instruct"  # Optional: specify a model
)

# Generate a response
response = llm.invoke("hi!")
print(response.content)

# Stream a response
llm.stream("Tell me a short story about a robot.")

# Get web Search results set web_search = True
llm.invoke("latest climate news",web_search = True)
llm.stream("Tell me a short story about a robot.",web_search = True)

Features

• Easy integration with LangChain
• WebSearch Hurray!!!
• Support for multiple HuggingChat models
• Built-in error handling and type checking

Configuration

You can configure the HuggingChat instance with the following parameters:

• hf_email: Your HuggingFace account email
• hf_password: Your HuggingFace account password
• model: (Optional) Specify a particular model to use from the available models list

Available Methods

• invoke: Generate a complete response for given input
• generate: Generate a ChatResult object (compatible with LangChain)
• stream: Stream the response as an iterator of message chunks
• pstream: Print the streamed response directly to console

Viewing Available Models

You can view the list of available models at any time using:

print(llm.get_available_models)

Error Handling

The package includes built-in error handling. If you encounter any issues during streaming or generation, informative error messages will be printed to the console.

Note on Credentials

Make sure to keep your HuggingFace credentials secure. You can set them as environment variables:

export HUGGINGFACE_EMAIL="[e-mail]"
export HUGGINGFACE_PASSWD="your_password"

Never share your credentials in public repositories or include them directly in your code.

License

This project is licensed under the MIT License.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Support

If you encounter any problems or have any questions, please open an issue on the GitHub repository.

Happy chatting with langchain_huggy!