Description:
We want to create a user-friendly chat-style GUI for running LLM inference via our Java engine (powered by TornadoVM), inspired by the layout shown in the attached screenshot.
Desired Features
-
Prompt Input
- A text box where users can enter prompts for the LLM
-
Dropdown Selectors
- Engine: Choose between TornadoVM, JVM, etc.
- Model: Select the model file to use for inference between Llama, Mistral etc
-
File Picker
- A "Browse" button to select .java files or model directories
-
Run Button
- A button to trigger inference (e.g., using tornado-llama-opencv)
-
Output Display
- A read-only area to show model responses and logs
- Optional: System Monitoring Panel
- Live GPU usage via nvtop or nvidia-smi
- Live CPU & memory stats via htop or Java system metrics
Preferably, integrated with the core code base with JavaFX, or external with QT or python QT.
Example POC:
