Llama-3-Instruct-8B-SPPO-Iter2 Benchmark Results

Llama-3-Instruct-8B-SPPO-Iter2 — benchmark results

Provider: Meta. Released 2024-06-25. Access: Open.

Unified ELO 1383 ± 30, rank #1301 of 1776 rated models, from 15 benchmark results.

Benchmark	Score	Metric	Percentile
Open LLM Leaderboard - IFEval	69.89	Score	82.8
Open Korean LLM Leaderboard	401.62	Average Score (%)	77.5
Open Chinese LLM - GSM8K	52.31	Accuracy (%)	71.8
Open Chinese LLM - TruthfulQA MC	55.54	Accuracy (%)	63.5
Open LLM Leaderboard - MMLU-Pro	29.91	Score	59.5
Open LLM Leaderboard - BBH	29.87	Score	51.2
Open Chinese LLM - CMMLU	53.31	Accuracy (%)	49
Open LLM Leaderboard - MATH Level 5	10.35	Score	48.5
Open Chinese LLM Leaderboard	56.42	Average Score (%)	46.9
Open Chinese LLM - C-Eval Semantic	68.56	Accuracy (%)	46
Open Chinese LLM - ARC Challenge	49.23	Accuracy (%)	38
Open Chinese LLM - WinoGrande	61.4	Accuracy (%)	26.9