Benchmark
BLINK - Semantic_Correspondence
Vision benchmarks measure how well models understand images. Scores are percentages.
All vision benchmarksScores from EvalAI data.
- Models measured
- 4
- Scores recorded
- 4
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 25%
- Mean
- 27.9%
- Median
- 3.6%
- Std dev
- 20.7%
- 10th percentile
- 29.3%
- 90th percentile
- 20.7%
- Min
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Qwen-VL MaxSemantic_Correspondence· EvalAI data | 29.3% |
| 2 | LLaVA 1.6 34BSemantic_Correspondence· EvalAI data | 27.9% |
| 3 | Gemini ProSemantic_Correspondence· EvalAI data | 22.1% |
| 4 | Claude 3 OpusSemantic_Correspondence· EvalAI data | 20.7% |