pythia-12B-sft-v8-rlhf-2k-steps: benchmark results
Provider: EleutherAI. Access: Open.
Aggregate Rating 1150 ± 16, rank #9973 of 10916 rated models, from 68 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - Abstract Algebra | 32 | Accuracy (%) | 68.1 |
| MMLU-by-task - Machine Learning | 33.93 | Accuracy (%) | 57.4 |
| MMLU-by-task - ARC Challenge | 40.44 | Accuracy (%) | 34.5 |
| MMLU-by-task - HellaSwag | 52.09 | Accuracy (%) | 34.2 |
| MMLU-by-task - Human Sexuality | 32.82 | Accuracy (%) | 34.2 |
| MMLU-by-task - Medical Genetics | 35 | Accuracy (%) | 34 |
| MMLU-by-task - World Religions | 35.09 | Accuracy (%) | 33.6 |
| MMLU-by-task - Jurisprudence | 30.56 | Accuracy (%) | 31.6 |
| MMLU-by-task - Marketing | 31.2 | Accuracy (%) | 31.1 |
| Open LLM Leaderboard v1 - GSM8K | 9.55 | Accuracy (%) (5-shot) | 30.2 |
| MMLU-by-task - Security Studies | 33.06 | Accuracy (%) | 30.1 |
| MMLU-by-task - College Biology | 27.78 | Accuracy (%) | 28.9 |
Interactive version: theaggregate.ai/model?slug=pythia-12b-sft-v8-rlhf-2k-steps · How It Works · Data refreshed daily, snapshot 2026-11-08.