Long context understanding on NIH Multi-needle
100AccuracyGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4ozero-shot=true2024.07 | 100 | |
| Llama 3 8Bzero-shot=true2024.07 | 98.8 | |
| Llama 3 405Bzero-shot=true2024.07 | 98.1 | |
| Llama 3 70Bzero-shot=true2024.07 | 97.5 | |
| Claude 3.5 Sonnetzero-shot=true2024.07 | 90.8 |