Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Diagram Understanding on AI2D lite
Loading...
82.8
Accuracy
PVM-8B (SFT + GRPO)
76.56
78.18
79.8
81.42
May 1, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
PVM-8B (SFT + GRPO)
Backbone=8B, Training...
2026.05
82.8
Euclid-8B
Backbone=8B, Training...
2026.05
82.6
PEARL-8B
Backbone=8B, Training...
2026.05
81.8
OneThinker-8B
Backbone=8B, Training...
2026.05
81.4
Qwen3-VL-8B (LoRA-SFT + GRPO)
Backbone=8B, Training...
2026.05
81
PVM-4B (SFT + GRPO)
Backbone=4B, Training...
2026.05
81
PVM-8B (SFT)
Backbone=8B, Training...
2026.05
80.8
PVM-4B (SFT)
Backbone=4B, Training...
2026.05
80
Qwen3-VL-8B-Instruct
Backbone=8B
2026.05
79.8
Qwen3-VL-8B (LoRA-SFT)
Backbone=8B, Training...
2026.05
79.8
CoMemo
Training Strategy=Visu...
2026.05
79.6
Qwen3-VL-8B (SFT + GRPO)
Backbone=8B, Training...
2026.05
79.6
Qwen3-VL-4B (LoRA-SFT)
Backbone=4B, Training...
2026.05
79.2
Qwen3-VL-8B (SFT)
Backbone=8B, Training...
2026.05
79
MemVR
Training Strategy=Visu...
2026.05
78.8
ICoT
Training Strategy=Visu...
2026.05
78.6
Qwen3-VL-4B (SFT + GRPO)
Backbone=4B, Training...
2026.05
78.6
Qwen3-VL-4B-Instruct
Backbone=4B
2026.05
78.4
Qwen3-VL-4B (SFT)
Backbone=4B, Training...
2026.05
77.6
Qwen3-VL-4B (LoRA-SFT + GRPO)
Backbone=4B, Training...
2026.05
76.8
Feedback
Search any
task
Search any
task