Share your thoughts, 1 month free Claude Pro on usSee more

Mathematical Reasoning on MathVista (Acc@1/Acc@4)

74.2Accuracy

Qwen2.5-VL-7B + DUPL

Updated 29d ago

Evaluation Results

Method	Links
Qwen2.5-VL-7B + DUPL 2025.10		74.2	-	-
PAPO-7B 2025.10		73.7	-	-
Qwen2.5-VL-7B + NoisyRollout 2025.10		72.6	-	-
MM-Eureka-Qwen-7B 2025.10		71.2	-	-
Qwen2.5-VL-7B + GRPO 2025.10		70.6	-	-
OpenVLThinker-7B 2025.10		70	-	-
Qwen2.5-VL-3B + DUPL 2025.10		69.4	-	-
VLAA-Thinker-7B 2025.10		68.3	-	-
Qwen2.5-VL-3B + NoisyRollout 2025.10		67.7	-	-
Qwen2.5-VL-7B 2025.10		67.2	-	-
CFPOG 2026.06		66.96	-	-
Coreset 2025.10		66.9	-	-
RECAP 2025.10		66.7	-	-
Qwen2.5-VL-3B + GRPO 2025.10		66.4	-	-
Uniform 2025.10		65.6	-	-
Reasoning-only 2025.10		65.5	-	-
R1-Onevision-7B 2025.10		63.9	-	-
VLAA-Thinker-7B 2025.10		63.9	-	-
LwF 2025.10		63.9	-	-
PropMix 2025.10		63.4	-	-
Vision-R1-7B 2025.10		61.8	-	-
Qwen2.5-VL-7B 2025.10		61.7	-	-
OpenVLThinker-7B 2025.10		59.1	-	-
Qwen2.5-VL-3B 2025.10		58.4	-	-
SFT + VIGIL 2026.06		56.6	-	-
PAPOG 2026.06		56.08	-	-
SFT + DA-DPO 2026.06		55.4	-	-
SFT + SimPO 2026.06		55.2	-	-
SFT only 2026.06		54.5	-	-
SFT + DPO 2026.06		54.1	-	-
SFT + HA-DPO 2026.06		53.8	-	-
GRPO 2026.06		53.58	-	-
SFT + VIGIL 2026.06		49.5	-	-
SFT + SimPO 2026.06		49.1	-	-
SFT + DA-DPO 2026.06		48.8	-	-
SFT only 2026.06		48.2	-	-
SFT + DPO 2026.06		48	-	-
GPT-5-Thinking* 2026.04		-	81.9	86.1
Gemini-2.5-Pro* 2026.04		-	80.9	85.2
Qwen2.5-VL-7B 2026.04		-	62.3	78.5
InternVL2.5-8B* 2026.04		-	68.2	72.8
R1-OneVision-7B 2026.04		-	64.1	69.4
VLAA-Thinker-7B 2026.04		-	68	70.8
Vision-R1-7B 2026.04		-	73.5	75.6
MUPO-Thinker-3B 2026.04		-	64.3	72.5
MUPO-Thinker-7B 2026.04		-	77.9	82.4