gui360-cooperative-lora-svd-sft – AI Model by Stevenshuqing | AlphaNeural AI

You can deploy this model and start earning money today!

Stevenshuqing/
gui360-cooperative-lora-svd-sft

like

0

cooperative-lora

Stevenshuqing/gui360-balanced

Qwen/Qwen2.5-VL-7B-Instruct

Views

No views yet

gui360-cooperative-lora-svd-sft

Cooperative LoRA (r=128, T=2 comm rounds) with SVD initialization from full SFT. TSR=18.6%, StepSR=65.3%. ~67M params. Starting checkpoint for RL.

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)	18.6%
StepSR (Step Success Rate)	65.3%
Progress	31.0%

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.