You can deploy this model and start earning money today!

namhokaist/
appgen-qwen3-vl-8b-sft-appgen-S-expanded-rare-policy-lora-r64

Name: appgen-qwen3-vl-8b-sft-appgen-S-expanded-rare-policy-lora-r64
Availability: InStock

Qwen/Qwen3-VL-8B-Instruct

finetune

apache-2.0

endpoints_compatible

Views

No views yet

AppGen Qwen3 policy SFT — arm S

This public checkpoint was trained only on user-owned synthetic AppGen data. It contains no AMEX, GUIOwl, AndroidWorld, ScreenSpot, RefExp, MoTIF, VWB, or other evaluation-benchmark examples.

Initialization: Qwen/Qwen3-VL-8B-Instruct at 0c351dd01ed87e9c1b53cbc748cba10e6187ff3b
Rows: 27,537
Dataset SHA-256: 1414f605e1e1dd4bfc9fc9daf548a40b197f17a51cb851d72021e7bdf4f55d09
Training type: lora
Learning rate: 1e-05
Optimizer steps: 861
Vision encoder and aligner: frozen
Global batch size: 32
Prompt: strict one-action <think> + <tool_call> policy with ten-step semantic history

For evaluation, use the exact appgen_system_prompt.txt included in this repository, enable explicit thinking, and provide normalized integer coordinates in [0, 999].

namhokaist/appgen-qwen3-vl-8b-sft-appgen-S-expanded-rare-policy-lora-r64

AppGen Qwen3 policy SFT — arm S

namhokaist/
appgen-qwen3-vl-8b-sft-appgen-S-expanded-rare-policy-lora-r64