Unverified50% confidenceBenchmarkExact time
经验内化微调把Qwen3-8B-27B的一次性得分从37.07提升到48.22，同时生成Token从641万降到57万（减少11倍）
1
Sources
50%
Confidence
Long-term
Relevance
10/2/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在终端基准（Terminal Bench）测试中，Qwen 3.6 27B得分约60.7，而Muse Glimmer只有51.7，明显落后76% similarUnverified微调后的Qwen3.5-4B模型在BU Bench V1基准上准确率从15%提升至53%74% similarUnverifiedQwen 3.6 Max在中性模式下端到端准确率仅12.3%72% similarUnverifiedQwen3 27B的IQ4XS量化版本性能相比原始满血版本仅下降约6%71% similarUnverifiedQwen3.7 Max的输出质量高度依赖提示词质量，同一模型在不同提示词下的性能差异可高达30%-50%71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/968728API
curl https://kongchang.com/api/v1/knowledge/claims/968728MCP
get_claim(id=968728)