Infinity-Instruct-7M-Gen-Llama3.1-8B: benchmark results
BAAI's SFT-only fine-tune of Llama 3.1 8B on the Infinity-Instruct 7M and Gen datasets, tuned without RLHF. Provider: BAAI. Released 2024-08-02. Access: Open.
Aggregate Rating 1480 ± 16, rank #851 of 1656 rated models, from 30 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-GPQA-Diamond | 28.79 | Normalized accuracy (%) | 98.1 |
| Open Korean LLM - Ko-IFEval | 51.65 | Strict accuracy, prompt/instruction mean (%) | 94.7 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 59.97 | Exact match, flexible extract (%) | 93.8 |
| Open Korean LLM Leaderboard | 46.39 | Average Score (%) | 93.2 |
| Open Ko-LLM Leaderboard | 46.39 | Average (%) | 93 |
| Open Korean LLM - Ko-EQ-Bench | 40.47 | EQ-Bench score | 91.7 |
| Open Portuguese LLM - ASSIN2 RTE | 93.54 | Macro F1 (%) | 90 |
| Open Korean LLM - KorNAT-Helpful | 49.23 | Normalized accuracy (%) | 77.8 |
| Open Portuguese LLM - ENEM | 70.82 | Accuracy (%) | 77.8 |
| Open Chinese LLM - GSM8K | 55.34 | Accuracy (%) | 76.2 |
| Open Portuguese LLM - BLUEX | 58.83 | Accuracy (%) | 74.9 |
| AlpacaEval 2.0 | 33.92 | LC Win Rate (%) | 73.4 |
Interactive version: theaggregate.ai/model?slug=infinity-instruct-7m-gen-llama3-1-8b · How It Works · Data refreshed daily, snapshot 2026-10-20.