List your company

Search on HR-MM Search

25.9Pass@1SFT + AXPO
7.49212.27117.0521.829May 27, 2026

Evaluation results

MethodLinks
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0525.9
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0524.4
SFTBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0523
BaseBackbone=Qwen3-VL-Thinking, Model Size=32B2026.0522.8
BaseBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0521
SFTBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0520.6
GRPOBackbone=Qwen3-VL-Thinking, Model Size=8B2026.0520.6
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0520.1
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0520
GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0518.4
SFT + AXPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0518.1
BaseBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0516.9
SFT + GRPOBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0515.7
SFTBackbone=Qwen3-VL-Thinking, Model Size=2B2026.0514.5
GRPOBackbone=Qwen3-VL-Thinking, Model Size=4B2026.0512.7
BaseBackbone=Qwen3-VL-Thinking, Model Size=2B2026.058.2