Do you think a few Qwen3.8-27B models working together could score as well as Fable-5 on LiveCodeBench Hard?
英文标题：Do you think a few Qwen3.8-27B models working together could score as well as Fable-5 on LiveCodeBench Hard?
Has anyone tested this? Ensemble of small Qwen models claiming Fable 5-level coding performance. A new paper claims that running several Qwen3.8-27B models together matches Fable 5’s accuracy on LiveCodeBench. The authors also say their setup paired with GPT Terra reaches Fable 5-level coding
打开原文相关 2 · 机会 2
中文
Has anyone tested this? Ensemble of small Qwen models claiming Fable 5-level coding performance. A new paper claims that running several Qwen3.8-27B models together matches Fable 5’s accuracy on LiveCodeBench. The authors also say their setup paired with GPT Terra reaches Fable 5-level coding
中文全文翻译正在补齐，可先查看下方英文原文。
English / 原文
Has anyone tested this? Ensemble of small Qwen models claiming Fable 5-level coding performance. A new paper claims that running several Qwen3.8-27B models together matches Fable 5’s accuracy on LiveCodeBench. The authors also say their setup paired with GPT Terra reaches Fable 5-level coding