internlm-20B Chat: benchmark results
Shanghai AI Lab's InternLM-20B Chat, pretrained on over 2.3 trillion tokens of English, Chinese and code data then aligned with SFT and RLHF, using a 60-layer deep architecture. Provider: Shanghai AI Lab. Released 2023-09-20. Access: Open.
Aggregate Rating 1350 ± 14, rank #7721 of 10916 rated models, from 75 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - College Physics | 42.16 | Accuracy (%) | 98.8 |
| MMLU-by-task - High School Computer Science | 66 | Accuracy (%) | 91.1 |
| MMLU-by-task - Electrical Engineering | 55.86 | Accuracy (%) | 90.8 |
| MMLU-by-task - Business Ethics | 63 | Accuracy (%) | 90.7 |
| MMLU-by-task - Anatomy | 54.07 | Accuracy (%) | 90 |
| MMLU-by-task - College Computer Science | 51 | Accuracy (%) | 89.8 |
| MMLU-by-task - High School European History | 76.36 | Accuracy (%) | 89.8 |
| MMLU-by-task - College Mathematics | 38 | Accuracy (%) | 89.5 |
| SALAD-Bench Base | 96.81 | Safety Score (%) | 89.4 |
| MMLU-by-task - Security Studies | 68.98 | Accuracy (%) | 89.3 |
| MMLU-by-task - Elementary Mathematics | 37.83 | Accuracy (%) | 89.1 |
| MMLU-by-task - High School Government And Politics | 85.49 | Accuracy (%) | 88.9 |
Interactive version: theaggregate.ai/model?slug=internlm-20b-chat · How It Works · Data refreshed daily, snapshot 2026-11-08.