Faro-Yi-9B: benchmark results
wenbopan's bilingual English-Chinese fine-tune of Yi 9B 200K on the Fusang-V1 dataset, focused on practicality and long-context use. Provider: wenbopan. Released 2024-03-27. Access: Open.
Aggregate Rating 1397 ± 16, rank #1234 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 70.77 | Accuracy (%) (5-shot) | 94.3 |
| Open Chinese LLM - C-Eval Semantic | 86.51 | Accuracy (%) | 80.2 |
| Open LLM Leaderboard v1 - GSM8K | 63 | Accuracy (%) (5-shot) | 76.3 |
| Open Chinese LLM - WinoGrande | 65.82 | Accuracy (%) | 74.9 |
| Open Chinese LLM Leaderboard | 62.85 | Official Nonzero-Task Mean (%) | 74.6 |
| Open Chinese LLM - CMMLU | 62.96 | Accuracy (%) | 72.5 |
| Open Chinese LLM - GSM8K | 51.78 | Accuracy (%) | 70.9 |
| Open Portuguese LLM - OAB Exams | 48.56 | Accuracy (%) | 70 |
| Open Chinese LLM - HellaSwag | 62.64 | Accuracy (%) | 68.9 |
| Open Chinese LLM - ARC Challenge | 55.12 | Accuracy (%) | 67.8 |
| Open Korean LLM - Ko-EQ-Bench | 21.49 | EQ-Bench score | 66.8 |
| Open Portuguese LLM - ENEM | 66.55 | Accuracy (%) | 64.4 |
Interactive version: theaggregate.ai/model?slug=faro-yi-9b · How It Works · Data refreshed daily, snapshot 2026-10-20.