Awqward2.5-32B-Instruct Benchmark Results

Awqward2.5-32B-Instruct — benchmark results

maldv's community merge of Qwen 2.5 32B fine-tunes (full-precision despite the AWQ pun). Provider: Other. Released 2024-12-18. Access: Open.

Unified ELO 1588 ± 16, rank #506 of 1776 rated models, from 179 benchmark results.

Benchmark	Score	Metric	Percentile
Open LLM Leaderboard - MATH Level 5	62.31	Score	99.9
Open Arabic LLM - Arabic MMLU HT High School Computer Science	86	Accuracy (%)	99.1
Open LLM Leaderboard - IFEval	82.55	Score	99.1
Open LLM Leaderboard - MMLU-Pro	52.48	Score	99.1
Open LLM Leaderboard - BBH	57.21	Score	98.8
Open Arabic LLM - Arabic MMLU Computer Science (University)	81.25	Accuracy (%)	97.8
Open Japanese LLM - Wiki PAS SET F1	11.86	Score (%)	97.4
Open Arabic LLM - Aratrust MentalHealth	96.05	Accuracy (%)	97.2
Open Japanese LLM - Jnli Exact Match	89.93	Score (%)	97.1
Open Arabic LLM - Alghafa Multiple Choice Grounded Statement Soqal Task	94.67	Accuracy (%)	96.9
Open Arabic LLM - Arabic MMLU HT Moral Scenarios	56.98	Accuracy (%)	96
Open Japanese LLM - CG	64.26	Score (%)	95.8