List your company

Multimodal Reasoning on COMT IID (Creation, Deletion, Selection, Update, Avg.)

89Creation ScoreZero-shot
50.5260.5170.580.49Dec 5, 2025

Evaluation results

MethodLinks
Zero-shotBackbone=Qwen3-VL-8B, Paradigm=Direct Ans., Training Stage=Standard2025.128928102137
Direct-FTBackbone=Qwen3-VL-8B, Paradigm=Direct Ans., Training Stage=Standard2025.128967495364.5
ILVRBackbone=Qwen3-VL-8B, Paradigm=Interleaved, Training Stage=Stage 2: Latent Relaxation2025.128773606270.5
ILVRBackbone=Qwen3-VL-8B, Paradigm=Interleaved, Training Stage=Stage 1: Latent Alignment2025.128463575564.8
MirageBackbone=Qwen3-VL-8B, Paradigm=Single-step, Training Stage=Stage 2: Latent Relaxation2025.128466545765.3
CoT-FTBackbone=Qwen3-VL-8B, Paradigm=Text CoT, Training Stage=Standard2025.128362494459.8
MirageBackbone=Qwen3-VL-8B, Paradigm=Single-step, Training Stage=Stage 1: Latent Alignment2025.128158435058
CoT-FTBackbone=Qwen2.5-VL-7B, Paradigm=Text CoT, Training Stage=Standard2025.128052454655.8
ILVRBackbone=Qwen2.5-VL-7B, Paradigm=Interleaved, Training Stage=Stage 2: Latent Relaxation2025.127168535160.8
ILVRBackbone=Qwen2.5-VL-7B, Paradigm=Interleaved, Training Stage=Stage 1: Latent Alignment2025.126966464757
Zero-shotBackbone=Qwen2.5-VL-7B, Paradigm=Direct Ans., Training Stage=Standard2025.126838351438.8
MirageBackbone=Qwen2.5-VL-7B, Paradigm=Single-step, Training Stage=Stage 2: Latent Relaxation2025.126562475056
MirageBackbone=Qwen2.5-VL-7B, Paradigm=Single-step, Training Stage=Stage 1: Latent Alignment2025.125354454248.5
Direct-FTBackbone=Qwen2.5-VL-7B, Paradigm=Direct Ans., Training Stage=Standard2025.125260514953