如果你经常在评测结果出来前纠结选哪个检查点,AutoEval 能提前帮你判断,分差够大时几乎不出错。
LMSYS Arena 推出 AutoEval,一种在完整评测结果出炉前比较候选模型检查点的早期信号。实验显示,当两个模型的 Arena 分数差≥10 分时,AutoEval 在超过 90% 的案例中正确识别出更高分模型;分数差≥15 分时,对所有测试对均正确排序。
AutoEval scores provide an early signal for comparing candidate model checkpoints before full live r...
AutoEval scores provide an early signal for comparing candidate model checkpoints before full live results become available. We compared AutoEval’s early ordering with the final live leaderboard. When two models were separated by at least 10 Arena points, AutoEval ranked the higher-scoring model correctly in more than 90% of cases. For gaps above 15 points, it correctly ordered every tested pair. 💬 1 🔄 1 ❤️ 2 👀 1190 📊 1 ⚡