Qwen2.5-VL-7B-Instruct-game-ui-navigator
Model Description
Fine-tuned from Qwen/Qwen2.5-VL-7B-Instruct using QLoRA (4-bit) with supervised fine-tuning.
This is a vision-language model trained to navigate game UI menus by looking at screenshots and deciding what to click or press to move forward.
Training Details
- Dataset: Game UI screenshots with navigation annotations
- LoRA rank: 16, alpha: 32
- Epochs: 3, Learning rate: 0.0002
Intended Use
Given a game UI screenshot, the model describes what it sees on screen and provides step-by-step instructions for navigating through menus to start the game.
Limitations
This model has only been trained on a small set of game UI screenshots and may not generalize well to unseen games or complex menu hierarchies.