pythia-12B-pre-v8-12.5k-steps: benchmark results
Provider: EleutherAI. Access: Open.
Aggregate Rating 1156 ± 16, rank #9879 of 10916 rated models, from 68 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - Abstract Algebra | 36 | Accuracy (%) | 91.7 |
| MMLU-by-task - High School Mathematics | 26.3 | Accuracy (%) | 40 |
| MMLU-by-task - Professional Law | 29.2 | Accuracy (%) | 35.9 |
| MMLU-by-task - Medical Genetics | 36 | Accuracy (%) | 34.7 |
| MMLU-by-task - Computer Security | 36 | Accuracy (%) | 33 |
| MMLU-by-task - Philosophy | 32.48 | Accuracy (%) | 33 |
| MMLU-by-task - HellaSwag | 51.1 | Accuracy (%) | 32.9 |
| MMLU-by-task - Moral Disputes | 30.64 | Accuracy (%) | 32.9 |
| MMLU-by-task - College Mathematics | 29 | Accuracy (%) | 32.7 |
| MMLU-by-task - College Medicine | 30.06 | Accuracy (%) | 32.7 |
| MMLU-by-task - Human Sexuality | 30.53 | Accuracy (%) | 32.7 |
| MMLU-by-task - World Religions | 33.92 | Accuracy (%) | 32.5 |
Interactive version: theaggregate.ai/model?slug=pythia-12b-pre-v8-12-5k-steps · How It Works · Data refreshed daily, snapshot 2026-11-08.