SOTA Text-based Visual Question Answering on TextVQA and PapersWithCode

73.78TextVQA Accuracy

AsymVLM

Updated 5d ago

Evaluation Results

Method	Links
AsymVLM 2026.05		73.78
SparseVLM 2026.05		73.66
Baseline 2026.05		73.54
FastV 2026.05		73.48
AsymVLM 2026.05		73.45
SparseVLM 2026.05		73.34
Uni-DPO 2025.06		73.23
AsymVLM 2026.05		73.2
SparseVLM 2026.05		73.2
FastV 2026.05		72.66
FastV 2026.05		72.24
Qwen2.5-VL-3B 2026.05		68.2
Cambrian-4B + Gaze attention 2026.05		66.2
Qwen2.5-VL-3B + InfiniPot-V 2026.05		66.1
Cambrian-4B + Gaze attention 2026.05		66
LLaVA-Next-LLaMA-3-8B 2026.05		64.7
Cambrian-4B + Gaze attention 2026.05		64.6
Cambrian-4B 2026.05		64.2
LLaVA-Next-Qwen2-7B 2026.05		63.6
Cambrian-4B + TokenCompre. + Gaze attention 2026.05		62.7
Cambrian-8B-CLIP-L 2026.05		62.4
Cambrian-4B + TokenCompre. + Gaze attention 2026.05		62.3
SimPO 2025.06		60.63
Qwen2.5-VL-3B + InfiniPot-V 2026.05		60.2
Cambrian-4B + TokenCompre. 2026.05		60.1
LLaVA-OV-7B + HERMES 2026.05		58
DPO 2025.06		52.73
BLIP-2-7B 2026.05		50.9
LLaVA-OV-7B + HERMES 2026.05		50.5
Baseline 2025.06		50.21
Cambrian-8B-DINOv2-L 2026.05		47.9
LLaVA-OV-7B 2026.05		44.2
SPHINX-7B 2026.05		38.3