Unverified50% confidenceBenchmarkExact time

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%

1

Sources

50%

Confidence

Long-term

Relevance

9/7/2026

First Seen

Sources

微调4B小模型：浏览器任务准确率从22%飙升至63%

Related Entities

Qwen3.5-4Bmodel BU Bench V1concept

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75% similar UnverifiedQwen 3.6 Max在中性模式下端到端准确率仅12.3%73% similar VerifiedQwen3.5的参数激活率为4.3%，而DeepSeek V3的671B模型激活率为5.5%72% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/869643

API

curl https://kongchang.com/api/v1/knowledge/claims/869643

MCP

get_claim(id=869643)