LLM-Based Chat Application

This is a language model-based chat application that provides users with relevant, up-to-date, and easy-to-understand responses by leveraging web search and a powerful language model. The basic flow involves performing a web search using DuckDuckGo, followed by generating a response using the Qwen2.5-0.5b-Instruct language model.

Features

Web Search: Performs real-time searches using DuckDuckGo to retrieve the latest relevant information.
Language Model Integration: Uses the qwen2.5-0.5b-instruct model for generating accurate and coherent responses.
Simple Interface: Built with Gradio to offer an easy-to-use web interface for users to interact with the model.
Efficient and Scalable: Optimized for performance using Intel-specific optimizations via optimum-intel and other acceleration libraries.

Dependencies

The following dependencies are required for running this application:

openvino>=2024.2.0
openvino-tokenizers[transformers]
torch>=2.1
datasets
duckduckgo-search
langchain-community
accelerate
gradio>=4.19
onnx<=1.16.1 (For Windows platform sys_platform=='win32')
einops
transformers>=4.43.1
transformers_stream_generator
tiktoken
bitsandbytes
optimum-intel (installed via git+https://github.com/huggingface/optimum-intel.git)
nncf (installed via git+https://github.com/openvinotoolkit/nncf.git)

Installation

To install the required dependencies, you can use the following command:

pip install -r requirements.txt

title: Llm Chatbot emoji: 👀 colorFrom: indigo colorTo: red sdk: gradio sdk_version: 5.4.0 app_file: app.py pinned: false

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference