Science Question Answering on ARC-Challenge Chinese
35.67ScoreInfLLM v2
Evaluation Results
| Method | Links | |
|---|---|---|
| InfLLM v2Training Tokens=100B, Inference Top-K=N/A, Training Top-K=22026.01 | 35.67 | |
| InfLLM v2Training Tokens=100B, Inference Top-K=2, Training Top-K=22026.01 | 35.41 | |
| InfLLM v2Training Tokens=100B, Inference Top-K=4, Training Top-K=42026.01 | 35.07 | |
| InfLLM v2Training Tokens=100B, Inference Top-K=N/A, Training Top-K=42026.01 | 34.98 | |
| PHSATraining Tokens=100B, Inference Top-K=N/A, Training Top-K=42026.01 | 34.39 | |
| DenseTraining Tokens=100B, Inference Top-K=N/A2026.01 | 34.3 | |
| DenseTraining Tokens=100B, Inference Top-K=22026.01 | 34.22 | |
| PHSATraining Tokens=100B, Inference Top-K=4, Training Top-K=42026.01 | 34.04 | |
| PHSATraining Tokens=100B, Inference Top-K=N/A, Training Top-K=22026.01 | 33.02 | |
| PHSATraining Tokens=100B, Inference Top-K=2, Training Top-K=22026.01 | 32.68 |