Visual Question Answering on GQA v1.0 (test)
63.3AccuracyShareGPT4V
Evaluation Results
| Method | Links | |
|---|---|---|
| ShareGPT4VLLM=Vicuna-7B, #Samples=1.9M2026.03 | 63.3 | |
| MobileVLM V2 7BLLM=Vicuna-7B, #Samples=3.6M2026.03 | 62.6 | |
| LLaVA-1.5LLM=Vicuna-7B, #Samples=1.2M2026.03 | 62 | |
| LLaVA-1.5-7BBackbone=LLaVA-1.5-7B, Retained Tokens=5762025.09 | 61.9 | |
| MoE-LLaVA-2BLLM=Qwen-1.5-1.8B, #Samples=2.2M2026.03 | 61.5 | |
| Cosine KD w/ Beta-KD (Instance)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 61.2 | |
| Align-KD w/ Beta-KD (Instance)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 60.8 | |
| Cosine KD w/ Beta-KD (Task)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 60.8 | |
| MoE-LLaVA-1.6BLLM=StableLM-1.6B, #Samples=2.2M2026.03 | 60.4 | |
| Align-KD w/ Beta-KD (Task)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 60.2 | |
| ZOO-PruneBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 60.03 | |
| Align-KDLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 60 | |
| DivPruneBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 59.97 | |
| Cosine KDLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 59.9 | |
| ZOO-PruneBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 59.49 | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 59.3 | |
| DivPruneBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 59.25 | |
| MobileVLM 3BLLM=MobileLLaMA 2.7B, #Samples=3.6M2026.03 | 59 | |
| ZOO-PruneBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 58.47 | |
| DivPruneBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 57.78 | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 57.6 | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 57.6 | |
| MobileVLM 1.7BLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 56.1 | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 56 | |
| LLAVADILLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 55.4 | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 55.1 | |
| MobileVLM -V2 1.7BLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 55.1 | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 52.7 | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 52.7 | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 49.6 | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 46.1 |