Bofeee5675/TongUI-3B
VISIONPricing:Input $0.32 / Cached $0.016 / Output $1.6Concurrent Unit Cost:1Model Size:3BQuant:BF16Context Size:32kPublished:Apr 18, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold
Bofeee5675/TongUI-3B is a 3 billion parameter multimodal language model, fine-tuned from Qwen/Qwen2.5-VL-3B-Instruct. It is specifically trained on the GUI-Net Dataset, which comprises internet-scale trajectories from multimodal web tutorials. This model is designed for generalized GUI agents, excelling at understanding and interacting with graphical user interfaces.
Loading preview...
TongUI-3B: Multimodal GUI Agent Model
TongUI-3B is a 3 billion parameter multimodal language model developed by Bofeee5675. It is built upon the Qwen/Qwen2.5-VL-3B-Instruct base model and has been fine-tuned using LoRA (Low-Rank Adaptation).
Key Capabilities
- Multimodal Understanding: Processes both visual and textual information, crucial for GUI interaction.
- GUI Agent Specialization: Specifically trained to understand and navigate graphical user interfaces.
- Internet-Scale Trajectory Learning: Leverages the extensive GUI-Net Dataset, which contains trajectories extracted from multimodal web tutorials, enabling robust learning of user interaction patterns.
Use Cases
- Automated GUI Interaction: Ideal for developing agents that can interact with web applications and software interfaces.
- Task Automation: Can be applied to automate complex workflows that involve navigating and manipulating graphical elements.
- Instruction Following: Excels at interpreting instructions related to GUI operations and executing them effectively.
For more detailed information, refer to the project page.