dolphin-2.9.2-Phi-3-Medium-abliterated: benchmark results
Cognitive Computations' Dolphin 2.9.2 tune of Phi-3 Medium 14B with refusals further ablated by FailSpy, for uncensored instruction, chat, and coding use. Provider: Cognitive Computations. Released 2024-06-03. Access: Open.
Aggregate Rating 1393 ± 17, rank #1251 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - ASSIN2 RTE | 93.63 | Macro F1 (%) | 90.8 |
| Open LLM Leaderboard - BBH | 63.83 | Score | 88.1 |
| Open LLM Leaderboard - MMLU-Pro | 45.25 | Score | 84.2 |
| Open Portuguese LLM - FaQuAD NLI | 78.14 | Macro F1 (%) | 82.8 |
| Open LLM Leaderboard - GPQA | 32.89 | Score | 80.1 |
| Open Chinese LLM - GSM8K | 56.33 | Accuracy (%) | 78 |
| Open Portuguese LLM - BLUEX | 59.81 | Accuracy (%) | 76.9 |
| Open Portuguese LLM - ENEM | 70.4 | Accuracy (%) | 76.2 |
| Open LLM Leaderboard - MuSR | 43.49 | Score | 74.2 |
| Open Portuguese LLM - OAB Exams | 48.7 | Accuracy (%) | 70.9 |
| Open Chinese LLM - C-Eval Semantic | 78.18 | Accuracy (%) | 67.8 |
| Open LLM Leaderboard - MATH Level 5 | 18.2 | Score | 67.8 |
Interactive version: theaggregate.ai/model?slug=dolphin-2-9-2-phi-3-medium-abliterated · How It Works · Data refreshed daily, snapshot 2026-10-20.