chat-cli

Chat with Groq-powered AI models directly in your terminal CLI.

Purpose

This chat-cli is designed to provide a simple, fast, and free alternative for chatting with Groq-powered AI models directly in your terminal. Unlike other large language models (LLMs) such as OpenAI's GPT, Gemini, or Claude, this tool is completely free to use--you don't pay anything-- and prioritizes ease of use.

The tool supports multiple models such as mixtral, llama-70b, and llama-8b-instant (default), and allows seamless interaction with these models via a command-line interface.

Features

• Model Selection: Easily switch between available models (e.g. mixtral, llama-70b, and llama-8b-instant) using the GROQ_MODEL environment variable.

• Verbose Mode: Enable detailed metrics such as elapsed time, tokens processed, and speed using the -v flag.

• Multiline Input: Type or paste multiline input seamlessly using the paste command.

• Clear Screen: Use the clear command to reset the terminal screen.

• Color-Coded Output:

  • User input is displayed in green (👤 You:).
  • Assistant output is displayed in magenta (🤖 Assistant:).

Installation

You can install the chat cli application using the go install command:

go install github.com/theDataFixer/chat-cli@latest

Usage

Environment Variables

If you haven't created your Groq API key, go to the GroqCloud Console, create an account and create the key.

Set your Groq API Key for authentication:

export GROQ_API_KEY=your-api-key-here

Set your preferred model (default is llama-8b-instant)

export GROQ_MODEL=llama-8b-instant # Options: mixtral, llama-70b, llama-instant

Start Chatting

After installation, you can start the chat CLI by running:

chat-cli

Enable Verbose Mode

To enable verbose mode (some metrics like time elapsed, tokens, etc), use:

chat-cli -v

Multiline Input

To paste multiline input:

1. Type paste and press Enter.
2. Paste your text.
3. Type done on a new line to finish.

Clear Screen

To clear the screen during a session:

clear

Exit Chat

To exit the chat session:

exit

Model details

The following models are:

Model Name	Description	Context Length
mixtral	Best for creativity	32,768 tokens
llama-70b	Versatile but larger	8,192 tokens
llama-instant	Fastest	8,192 tokens

You can change the model dynamically by setting the GROQ_MODEL environment variable.

Troubleshooting

If you encounter issues during installation or running the application, check the following:

• Ensure Go is properly installed and configured.
• Verify that your GOPATH is set correctly.
• Make sure you have set your Groq API key (GROQ_API_KEY) in your environment variables.

Example Session

Chat started using model: llama-8b-instant (type 'exit' to quit)
Type 'clear' to clear the screen
For multiline input, type 'paste' and press Enter

👤 You: Tell me a palindrome Python code
🤖 Assistant: Here's a simple Python code to check if a given word is a palindrome:

def is_palindrome(word):
    return word == word[::-1]

# Test the function
print(is_palindrome("racecar"))  # True
print(is_palindrome("python"))   # False

👤 You: Thanks!
🤖 Assistant: You're welcome! Let me know if you need anything else.

Notes:

• It is a simple binary, with some minor tweaks regarding decorations.
• It is a project for learning Golang, so you may find some bad practices or some bad code (if so, please let me know).
• Hopefully this chat-cli will become more robust either by contributions or by personal upgrades.

Contributing

Contributions are welcome. Please fork the repository and submit a pull request with your changes.

Questions?

For any inquiries or support, please reach out at: [email protected]