Visual Question Answering On Mm Vet

Métriques

GPT-4 score

Résultats

Résultats de performance de divers modèles sur ce benchmark

Nom du modèle	GPT-4 score	Paper Title	Repository
SoM-LLaVA-1.5-T	37.2	List Items One by One: A New Data Source and Learning Paradigm for Multimodal LLMs
LLaVA-1.5-7B (VG-S)	40.4	ProVision: Programmatically Scaling Vision-centric Instruction Data for Multimodal Language Models	-
MMCTAgent (GPT-4 + GPT-4V)	74.24	MMCTAgent: Multi-modal Critical Thinking Agent Framework for Complex Visual Reasoning	-
DeepSeek-VL	41.5	DeepSeek-VL: Towards Real-World Vision-Language Understanding
VOLCANO 13B	38.0	Volcano: Mitigating Multimodal Hallucination through Self-Feedback Guided Revision
LOVA$^3$	35.2	LOVA3: Learning to Visual Question Answering, Asking and Assessment
Qwen2-VL-2B (finetuned on GAP-VQA train)	52.43	Gamified crowd-sourcing of high-quality data for visual fine-tuning	-
mPLUG-Owl2	36.3±0.1	mPLUG-Owl2: Revolutionizing Multi-modal Large Language Model with Modality Collaboration
LLaVA-Plus-7B (All Tools)	27.5±0.3	LLaVA-Plus: Learning to Use Tools for Creating Multimodal Agents
InternLM2+ViT (QMoSLoRA)	35.2	Mixture-of-Subspaces in Low-Rank Adaptation
JanusFlow	30.9	JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation
ConvLLaVA	45.9	ConvLLaVA: Hierarchical Backbones as Visual Encoder for Large Multimodal Models
Mini-Gemini	53.0	Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models
InfiMM-HD	38.9	InfiMM-HD: A Leap Forward in High-Resolution Multimodal Understanding	-
InternVL2-26B (SGP, token ratio 9%)	52.10	A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for Accelerating Large VLMs	-
LLaVA-1.5-LLaMA3-8B	37.8	What If We Recaption Billions of Web Images with LLaMA-3?	-
InternVL2-Llama3-76B	64.4	-	-
LLaVA-HR-X	35.5	Feast Your Eyes: Mixture-of-Resolution Adaptation for Multimodal Large Language Models
CoLLaVO	40.3	CoLLaVO: Crayon Large Language and Vision mOdel	-
Silkie	49.9	Silkie: Preference Distillation for Large Visual Language Models	-

0 of 229 row(s) selected.