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

Table Question Answering on TAT-QA

75.88Accuracy

Qwen3-VL-8B + TABQAWORLD

Updated 12d ago

Evaluation Results

Method
Qwen3-VL-8B + TABQAWORLD 2026.04	75.88	-
Qwen2.5-VL-7B + TABQAWORLD 2026.04	71.36	-
Qwen2.5-VL-Instruct + Vanilla SFT 2025.02	62.95	62.15
Qwen2.5-VL-Instruct + HIPPO 2025.02	62.82	62.9
TableDART (TG2-7B+Ovis2-8B) 2026.04	62.05	-
MiniCPM-V-2.6 + HIPPO 2025.02	60.75	65.46
HIPPO-8B 2026.04	60.75	-
Qwen3-VL-8B 2026.04	59.8	-
TableDART (TG2-7B+Qwen2.5-VL-7B) 2026.04	59.07	-
MiniCPM-V-2.6 + Vanilla SFT 2025.02	58.91	61.93
Qwen2.5-VL-Instruct 2025.02	56.73	58.58
MiniCPM-V-2.6 2025.02	52.46	62.12
SynTab-LLaVA-7B 2026.04	51.94	-
Qwen2.5-VL-7B 2026.04	51.94	-
MiniCPM-V-2.6 (Image Only) 2025.02	51.55	60.42
MiniCPM-V-2.6-8B 2026.04	51.55	-
Qwen2.5-VL-Instruct (Image Only) 2025.02	50.9	56.85
TableGPT2-7B (Text-only Path) 2026.04	50.39	-
Ovis2-8B (Image-only Path) 2026.04	47.67	-
Table-LLaVA + HIPPO 2025.02	43.26	55.52
Gemini 2.0 Flash 2026.04	35.62	-
Table-LLaVA 2025.02	27.84	50.98
GPT-4o (Image Only) 2025.02	25.12	54.26
GPT-4o 2025.02	21.24	55.58
TableLlama + Markdown 2025.02	19.04	37.86
TableLlama-7B 2026.04	19.04	-
Table-LLaVA (Image Only) 2025.02	15.67	38.09
Llama2 2025.02	13.73	17.53
Llama-2-7B 2026.04	13.73	-
Llama3-Instruct 2025.02	13.08	31.98
Llama3-Instruct-8B 2026.04	13.08	-
Table-LLaVA (Image Only) 2025.02	12.82	35.69
Table-LLaVA-7B 2026.04	12.82	-
Monkey 2025.02	12.31	14.16
Llama3.1-Instruct 2025.02	10.06	35.42
TabPedia 2025.02	9.59	20.84
Qwen2.5-Instruct 2025.02	9.27	61.54
Vary-toy 2025.02	8.81	5.77
Qwen2-72B-Instruct 2025.02	4.87	57.11
LLaVA v1.5 2025.02	2.97	9.82
mPLUG-Owl2 2025.02	0.39	7.76
InternLM-XComposer 2025.02	0.26	0.77
MiniGPT-4 2025.02	0.13	0.28
Qwen-VL 2025.02	0.13	0.83
mPLUG-Owl 2025.02	0.13	3.16
TableLlama + Oracle 2025.02	-	55.33