CarbonVillain-en-10.7B-v4 Benchmark Results

CarbonVillain-en-10.7B-v4 — benchmark results

Provider: Other. Released 2024-03-01. Access: Open.

Unified ELO 1392 ± 18, rank #1264 of 1776 rated models, from 25 benchmark results.

Benchmark	Score	Metric	Percentile
Open Chinese LLM - TruthfulQA MC	61.01	Accuracy (%)	85.5
Open Korean LLM Leaderboard	419.94	Average Score (%)	79.2
Open Medical LLM - MMLU Clinical Knowledge	73.58	Accuracy (%)	69.6
Open Medical LLM - MMLU College Medicine	65.32	Accuracy (%)	69.3
Open Medical LLM - MMLU Professional Medicine	72.79	Accuracy (%)	66.8
Open Chinese LLM - ARC Challenge	54.27	Accuracy (%)	64.7
Open LLM Leaderboard - GPQA	7.49	Score	63.1
Open LLM Leaderboard - BBH	31.81	Score	58.7
Open Chinese LLM Leaderboard	56.78	Average Score (%)	54.6
Open Medical LLM - MMLU College Biology	74.31	Accuracy (%)	52.8
Open LLM Leaderboard - IFEval	45.79	Score	51.3
Open Medical LLM - MedQA (USMLE)	54.83	Accuracy (%)	50.6