Temporal Relation Extraction On Vinoground

Metrics

Group Score

Text Score

Video Score

Results

Performance results of various models on this benchmark

Model Name	Group Score	Text Score	Video Score	Paper Title	Repository
GPT-4o	24.6	54	38.2	-	-
LLaVA-NeXT-Video-7B	6.2	21.8	25.6	-	-
LLaVA-NeXT-Video-7B (CoT)	6.8	21.8	26.2	-	-
LLaVA-NeXT-Video-34B	3.8	23	21.2	-	-
Qwen2-VL-7B	15.2	40.2	32.4	Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution	-
GPT-4o (CoT)	35	59.2	51	-	-
Phi-3.5-Vision	6.2	24	22.4	-	-
Claude 3.5 Sonnet	10.6	32.8	28.8	-	-
ImageBind	0.6	9.4	3.4	ImageBind: One Embedding Space To Bind Them All	-
LLaVA-OneVision-Qwen2-72B	21.8	48.4	35.2	LLaVA-OneVision: Easy Visual Task Transfer	-
VTimeLLM	5.2	19.4	27	VTimeLLM: Empower LLM to Grasp Video Moments	-
MA-LMM-Vicuna-7B	6.8	23.8	25.6	MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding	-
LLaVA-OneVision-Qwen2-7B	14.6	41.6	29.4	LLaVA-OneVision: Easy Visual Task Transfer	-
Gemini-1.5-Pro (CoT)	12.4	37	27.6	Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context	-
Gemini-1.5-Pro	10.2	35.8	22.6	Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context	-
InternLM-XC-2.5	9.6	28.8	27.8	InternLM-XComposer-2.5: A Versatile Large Vision Language Model Supporting Long-Contextual Input and Output	-
VideoCLIP	1.2	17	2.8	VideoCLIP: Contrastive Pre-training for Zero-shot Video-Text Understanding	-
Video-LLaVA-7B	6.6	24.8	25.8	Video-LLaVA: Learning United Visual Representation by Alignment Before Projection	-
LanguageBind	1.2	10.6	5	LanguageBind: Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignment	-
VideoLLaMA2-72B	8.4	36.2	21.8	VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs	-

0 of 24 row(s) selected.