Perception and Cognition Evaluation on MME Perception and Cognition v1.0
1,567.4MME^P ScoreShareGPT4V
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ShareGPT4VLLM=Vicuna-7B, #Samples=1.9M2026.03 | 1,567.4 | 78.4 | |
| MobileVLM V2 7BLLM=Vicuna-7B, #Samples=3.6M2026.03 | 1,559 | 78 | |
| LLaVA-1.5LLM=Vicuna-7B, #Samples=1.2M2026.03 | 1,510.7 | 75.5 | |
| Cosine KD w/ Beta-KD (Task)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,352 | 67.6 | |
| Cosine KD w/ Beta-KD (Instance)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,350.3 | 67.5 | |
| Align-KD w/ Beta-KD (Instance)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,343 | 67.1 | |
| Align-KD w/ Beta-KD (Task)LLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,315.1 | 65.8 | |
| Cosine KDLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,308.4 | 65.4 | |
| MoE-LLaVA-1.6BLLM=StableLM-1.6B, #Samples=2.2M2026.03 | 1,300 | 65 | |
| MoE-LLaVA-2BLLM=Qwen-1.5-1.8B, #Samples=2.2M2026.03 | 1,292 | 64.6 | |
| MobileVLM 3BLLM=MobileLLaMA 2.7B, #Samples=3.6M2026.03 | 1,288.9 | 64.4 | |
| Align-KDLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,288.9 | 64.4 | |
| MobileVLM -V2 1.7BLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,246.3 | 62.3 | |
| MobileVLM 1.7BLLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,196.2 | 59.8 | |
| LLAVADILLM=MobileLLaMA 1.4B, #Samples=3.6M2026.03 | 1,178.6 | 58.9 |