Introduction

Welcome to PacketLLM! This package provides a seamless way to interact with OpenAI’s Large Language Models (LLMs) such as GPT-4o directly within your RStudio environment. It runs as an RStudio Gadget, offering a familiar chat interface without the need to switch between applications.

With PacketLLM, you can: - Generate code and explanations. - Analyze text from uploaded files (.R, .pdf, .docx). - Manage multiple conversations in tabs. - Customize model behavior through settings.

This vignette will guide you through the essential setup and basic usage.

Prerequisites: OpenAI API Key

PacketLLM requires an OpenAI API key.

Obtain an API Key:
Sign up on the OpenAI Platform and create an API key.
Configure the API Key:
The recommended method is to set the OPENAI_API_KEY environment variable.

Using .Renviron (recommended):

Add the following line to your user or project .Renviron file (replace 'your_secret_api_key_here' with your actual key):
```
OPENAI_API_KEY='your_secret_api_key_here'
```
You can open the file using:
```
usethis::edit_r_environ()         # User-level
usethis::edit_r_environ("project")  # Project-level
```
Important: Restart your R session after editing .Renviron.

Temporary Solution:

For the current session only, run:
```
Sys.setenv(OPENAI_API_KEY = "your_secret_api_key_here")
```

Installation

Install PacketLLM from GitHub:

# Uncomment the next line if you need the remotes package
# install.packages("remotes")
remotes::install_github("AntoniCzolgowski/PacketLLM")

Launching the Chat Gadget

Once installed and your API key is set, launch the gadget from the R console:

library(PacketLLM)
run_llm_chat_app()

This command should be executed in an interactive RStudio session.

Understanding the Interface

The PacketLLM interface consists of:

Top Navigation Bar:
- Settings: Configure the model, temperature, and system message (before sending the first message).
- New Conversation: Create a new chat tab.
- Close App (X): Exit the PacketLLM gadget.
Conversation Tabs:
Switch between multiple chat sessions; each tab can be closed individually.
Chat History Area:
Displays the dialogue between you and the AI.
Input Area:
- + Button: Open a file browser to attach supported files (.R, .pdf, .docx) as context.
- Attachments Staging Area: Lists selected files pending for upload.
- Text Input Box: Where you type your message.
- Send Button: Send your message and any attached files to the model.

Basic Workflow

Launch the App: Run run_llm_chat_app().
Start Chatting: Type your question or prompt.
Send Message: Click the Send button.
Wait for Response: The app will show a processing indicator while waiting for the model’s reply.

Optional steps: - Add File Context: Click the + button to attach files before sending. - Start a New Conversation: Click New Conversation for a fresh session. - Adjust Settings: Use Settings to change model options before sending the first message.

Exploring Further

Experiment with different models, system prompts, and file attachments to enhance your workflow with PacketLLM. For issues or suggestions, please visit the GitHub Issues page.