Title

Custom Chatbot Langchain (FREE)

In case you don't have a powerful computer for GPU, use the colab link above, and make sure you are using the T4 GPU on Google Collab.

Check Out The Colab Project

Description

Make a bot that answers question from your txt file. Human like responses, and accurate using the llama 7B. Everything in this is free. I have added a short txt file that has some few stuffs about the university of Iceland, and can answer few questions based on that txt file.

Installation

1. Clone Repository:

   git clone https://github.com/Cyboghostginx/Custom-Chatbot-Langchain-FREE-.git
   cd Custom-Chatbot-Langchain-FREE-

2. Ollama Installation:

   curl https://ollama.ai/install.sh | sh
   ollama run llama2

3. Langchain Installation:

   pip install git+https://github.com/langchain-ai/langchain.git@8d6faf56657070137a16669bc2420a08a5ab7f24#subdirectory=libs/langchain

4. Tiktoken Installation:

   pip install tiktoken

5. FastEmbed Installation:

   pip install fastembed

6. Qdrant-Client Installation:

   pip install qdrant-client

Usage

python bot.py

Tested

I used a macbook pro m1, 16gb ram for this.

Splitting tx into chunks - Almost instantly
Embedding finished in 1.6 seconds
Similarity search - Almost instantly
Chat model response - 20 seconds for the the llama 7b model.

Note that when this is hosted on a better gpu, the chat model to would respond almost immediately. Thanks!