Infinity-Instruct-3M-0625-Yi-1.5-9B: benchmark results
BAAI's SFT-only fine-tune of Yi-1.5 9B, trained in two stages on the Infinity-Instruct 3M and 0625 datasets without RLHF (July 2024). Provider: BAAI. Released 2024-07-09. Access: Open.
Aggregate Rating 1429 ± 17, rank #1102 of 1656 rated models, from 30 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - GPQA | 35.4 | Score | 89.1 |
| Open Chinese LLM - GSM8K | 62.02 | Accuracy (%) | 87 |
| Open LLM Leaderboard - MuSR | 45.75 | Score | 86.8 |
| Open Chinese LLM Leaderboard | 65.64 | Official Nonzero-Task Mean (%) | 82.2 |
| Open Korean LLM - KorNAT-Helpful | 49.53 | Normalized accuracy (%) | 80.1 |
| Open Chinese LLM - C-Eval Semantic | 86.01 | Accuracy (%) | 78.1 |
| Open Portuguese LLM - FaQuAD NLI | 76.92 | Macro F1 (%) | 78 |
| Open Korean LLM - Ko-IFEval | 36.46 | Strict accuracy, prompt/instruction mean (%) | 75.5 |
| Open Chinese LLM - HellaSwag | 64.06 | Accuracy (%) | 75.4 |
| Open LLM Leaderboard - MMLU-Pro | 41.18 | Score | 74.9 |
| Open LLM Leaderboard - BBH | 55.09 | Score | 73.8 |
| Open Ko-LLM Leaderboard | 39.74 | Average (%) | 73.6 |
Interactive version: theaggregate.ai/model?slug=infinity-instruct-3m-0625-yi-1-5-9b · How It Works · Data refreshed daily, snapshot 2026-10-20.