Captioning on Personalization Benchmark (Single, Multi, Weight metrics)
83.5Single ScoreICPT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ICPTLVLM=Qwen3VL-8B2026.05 | 83.5 | 82.1 | 82.6 | |
| ICPTLVLM=InternVL3-8B2026.05 | 82.8 | 80.4 | 81.2 | |
| PLVMLVLM=InternVL3-8B2026.05 | 80.6 | 76.2 | 77.7 | |
| PLVMLVLM=Qwen3VL-8B2026.05 | 78.4 | 70.8 | 73.3 | |
| ICPTLVLM=LLaVA-NeXT-34B2026.05 | 76.2 | 61.4 | 66.3 | |
| PLVMLVLM=LLaVA-NeXT-34B2026.05 | 74 | 54.5 | 61 | |
| ICPTBackbone=LLaVA-NeXT-7B2026.05 | 71.3 | 62.4 | 65.4 | |
| MC-LLaVABackbone=LLaVA-NeXT-7B2026.05 | 68.6 | 53.8 | 58.7 | |
| GPT-4o + ICLBackbone=GPT-4o2026.05 | 67.4 | 60.8 | 63 | |
| PLVMBackbone=LLaVA-NeXT-7B2026.05 | 66.4 | 50.7 | 55.9 | |
| PeKitBackbone=LLaVA-NeXT-7B2026.05 | 65.8 | 48.6 | 54.3 | |
| ICLLVLM=Qwen3VL-8B2026.05 | 63.5 | 57.2 | 59.3 | |
| PVITBackbone=LLaVA-NeXT-7B2026.05 | 62.3 | 46.7 | 51.9 | |
| RAPBackbone=LLaVA-NeXT-7B2026.05 | 58 | 49.3 | 52.2 | |
| ICLLVLM=InternVL3-8B2026.05 | 57.5 | 53.8 | 55 | |
| Yo’ LLaVABackbone=LLaVA-NeXT-7B2026.05 | 50.6 | 38.5 | 42.5 | |
| MyVLMBackbone=LLaVA-NeXT-7B2026.05 | 49.3 | — | — | |
| ICLLVLM=LLaVA-NeXT-34B2026.05 | 32.8 | 21.5 | 25.3 | |
| VanillaLVLM=Qwen3VL-8B2026.05 | 30.6 | 25.7 | 27.3 | |
| ICLBackbone=LLaVA-NeXT-7B2026.05 | 28.9 | 13.7 | 18.8 | |
| VanillaLVLM=InternVL3-8B2026.05 | 23.5 | 26.3 | 25.4 | |
| VanillaLVLM=LLaVA-NeXT-34B2026.05 | 23.1 | 14.7 | 17.5 | |
| VanillaBackbone=LLaVA-NeXT-7B2026.05 | 20.8 | 15.9 | 17.5 |