精选理由
Muse Spark 1.2 (xHigh) 模型发布了，在处理命令行错误上提升明显，用户反馈也更好，但纠错和引导能力略有下降。
Muse Spark 1.2 (xHigh) 模型在 Bash Recovery 基准上提升了 11.4%。该模型在 Confirmed Success 指标上提升了 6%。Praise vs. Complaint 指标下降了 5.7%。Steerability 指标下降了 2.5%。Tool Hallucination 指标提升了 1.1%。
原文 · lmarena.ai
By signal, Muse Spark 1.2 (xHigh) landed: +11.4% in Bash Recovery (recovery from CLI errors) +6% i...
By signal, Muse Spark 1.2 (xHigh) landed: +11.4% in Bash Recovery (recovery from CLI errors) +6% in Confirmed Success ("yes, that worked" user feedback) -5.7% in Praise vs. Complaint (implicit user sentiment) -2.5% in Steerability (course-correction after pushback) No issues with Tool Hallucination (+1.1%) 💬 1 🔄 0 ❤️ 2 👀 1015 📊 1 ⚡