List your company
ResearchFormerly WizWand

Multimodal Question Answering on MMQA

70.5AccuracyOriginal Model
57.560.87564.2567.625Dec 28, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Original ModelBackbone=LLaVA-1.5-13B, Retention Ratio=1.02025.1270.5
TrimTokenator-LCBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1269
VisionZipBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1267.5
TrimTokenator-LCBackbone=Yi-VL-6B, Retention Ratio=0.22025.1267.5
Original ModelBackbone=LLaVA-1.5-7B, Retention Ratio=1.02025.1267
SparseVLMBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1267
TrimTokenatorBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1267
Original ModelBackbone=Yi-VL-6B, Retention Ratio=1.02025.1267
FastVBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1266.5
CATPBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1266.5
Original ModelBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1266
VTWBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1266
DivPruneBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.1266
DivPruneBackbone=Yi-VL-6B, Retention Ratio=0.22025.1266
VTWBackbone=Yi-VL-6B, Retention Ratio=0.22025.1265.5
TrimTokenatorBackbone=Yi-VL-6B, Retention Ratio=0.22025.1264.5
SparseVLMBackbone=Yi-VL-6B, Retention Ratio=0.22025.1264
CATPBackbone=Yi-VL-6B, Retention Ratio=0.22025.1263.5
TrimTokenator-LCBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1263
FastVBackbone=Yi-VL-6B, Retention Ratio=0.22025.1263
VisionZipBackbone=Yi-VL-6B, Retention Ratio=0.22025.1263
DivPruneBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1262.5
TrimTokenator-LCBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1262.5
TrimTokenatorBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1262
SparseVLMBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1262
DivPruneBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1262
TrimTokenatorBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1261.5
FastVBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1261
VisionZipBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1261
SparseVLMBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1260.5
VTWBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1260.5
CATPBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.1260.5
VTWBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1259.5
VisionZipBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1258.5
FastVBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1258
CATPBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.1258