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

Cross-modal Alignment ((I+A)→T) on Image-Text-Audio

29.97CLIP Score

UnifiedIO2-L

Updated 1mo ago

Evaluation Results

Method	Links
UnifiedIO2-L 2026.06		29.97	-
LLaVA-NeXT 2026.06		29.47	-
MUNI† 2026.06		29.18	-
FlowBind 2026.06		29.14	-
MMVAE 2026.06		29.14	-
MUNI 2026.06		29.05	-
MoPoE 2026.06		29.05	-
MoPoE 2026.06		27.81	36.02
MUNI† 2026.06		27.6	36.61
FlowBind 2026.06		27.54	39.56
OmniFlow 2026.06		27.48	-
MMVAE 2026.06		26.57	35.24
CoDi 2026.06		26.05	-
MUNI 2026.06		25.2	39.75
OmniFlow 2026.06		24.73	36.26
CoDi 2026.06		24.05	33.72
Qwen2-Audio 2026.06		-	45.45
UnifiedIO2-L 2026.06		-	35.48
CoDi 2026.06		-	32.72
OmniFlow 2026.06		-	44.74
FlowBind 2026.06		-	41.8
MUNI 2026.06		-	39.38
MUNI† 2026.06		-	39.45
MMVAE 2026.06		-	38.3
MoPoE 2026.06		-	38.98