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

Table Fact Verification on InfoTabs

81.62Accuracy

Qwen2-72B-Instruct

Updated 11d ago

Evaluation Results

Method
Qwen2-72B-Instruct 2025.02	81.62	57.11
Qwen3-VL-8B + TABQAWORLD 2026.04	81.41	-
GPT-4o 2025.02	78.46	55.58
Qwen2.5-Instruct 2025.02	78.29	61.54
Qwen2.5-VL-Instruct + HIPPO 2025.02	78.22	62.9
Qwen2.5-VL-Instruct + Vanilla SFT 2025.02	78.13	62.15
Qwen2.5-VL-7B + TABQAWORLD 2026.04	76.38	-
MiniCPM-V-2.6 + Vanilla SFT 2025.02	76.22	61.93
TableDART (TG2-7B+Ovis2-8B) 2026.04	76.22	-
GPT-4o (Image Only) 2025.02	76	54.26
Qwen3-VL-8B 2026.04	75.88	-
MiniCPM-V-2.6 + HIPPO 2025.02	75.74	65.46
HIPPO-8B 2026.04	75.74	-
Qwen2.5-VL-Instruct 2025.02	75.12	58.58
Ovis2-8B (Image-only Path) 2026.04	74.11	-
Qwen2.5-VL-Instruct (Image Only) 2025.02	73.46	56.85
MiniCPM-V-2.6 2025.02	73.18	62.12
MiniCPM-V-2.6-8B 2026.04	73.03	-
MiniCPM-V-2.6 (Image Only) 2025.02	72.03	60.42
TableDART (TG2-7B+Qwen2.5-VL-7B) 2026.04	71.46	-
TableGPT2-7B (Text-only Path) 2026.04	71.07	-
Qwen2.5-VL-7B 2026.04	70.13	-
SynTab-LLaVA-7B 2026.04	69.42	-
Table-LLaVA + HIPPO 2025.02	68.71	55.52
Llama3.1-Instruct 2025.02	68.44	35.42
Table-LLaVA (Image Only) 2025.02	66.91	38.09
Table-LLaVA (Image Only) 2025.02	65.26	35.69
Table-LLaVA-7B 2026.04	65.26	-
Table-LLaVA 2025.02	63.63	50.98
Gemini 2.0 Flash 2026.04	54.31	-
Llama3-Instruct 2025.02	54	31.98
Llama3-Instruct-8B 2026.04	54	-
TableLlama + Markdown 2025.02	46.57	37.86
TableLlama-7B 2026.04	46.57	-
Llama2 2025.02	38.92	17.53
Llama-2-7B 2026.04	38.92	-
TabPedia 2025.02	34.7	20.84
LLaVA v1.5 2025.02	28.31	9.82
mPLUG-Owl2 2025.02	26.19	7.76
Monkey 2025.02	22.11	14.16
Vary-toy 2025.02	6.98	5.77
mPLUG-Owl 2025.02	5.53	3.16
InternLM-XComposer 2025.02	1.11	0.77
Qwen-VL 2025.02	0.65	0.83
MiniGPT-4 2025.02	0.1	0.28
TableLlama + Oracle 2025.02	-	55.33