gui360-cooperative-lora-rl-step25
Cooperative LoRA after RL (GSPO) training. TSR=20.8%, StepSR=67.9%. Best LoRA result, 94% of full SFT performance with ~142M params.
Base Model
- Qwen2.5-VL-7B-Instruct
Training Data
- GUI-360 balanced 2K episodes (17,264 steps)
- Action types: click, type, swipe (balanced sampling)
Evaluation (GUI-360 test 1K balanced)
| Metric | Value |
|---|---|
| TSR (Task Success Rate) | 20.8% |
| StepSR (Step Success Rate) | 67.9% |
| Progress | 34.5% |
Full Ranking
| # | Method | TSR | Params |
|---|---|---|---|
| 1 | Full-param SFT step-250 | 22.2% | 7.6B |
| 2 | V15 Cooperative RL step-25 | 20.8% | ~142M |
| 3 | PEFT Cooperative r=128 (SVD) | 18.6% | ~67M |
| 4 | PEFT Standard r=128 (SVD) | 18.1% | ~67M |
| 5 | Base model (zero-shot) | 2.4% | — |
Citation
Part of the Cooperative LoRA research for GUI agents.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Stevenshuqing/gui360-cooperative-lora-rl-step25
Base model
Qwen/Qwen2.5-VL-7B-Instruct