RCreddit.com
23
·开发者社区 · RSS
<15B local models to test on medical benchmark?
这条记录涉及生成能力或端侧推理进展，适合跟踪模型效率、部署门槛和应用机会。
一位用户计划在未来几周内，使用意大利居民的医学考试来测试其笔记本电脑上运行的小型模型。他希望通过这种非标准基准测试，了解这些模型与普通住院医师相比的表现。去年，一些32B模型已经在此考试中达到了最高分，因此他认为测试小型模型是可行的，并正在寻求模型推荐。
HI, in a few weeks the medical exam for italian residents will drop, and I would like to compare small models that run on my laptop against it. It would not be a standard benchmark for llms but it would give a sense on how much better does it perform compared to the average medical resident. Do you have any reccomendation? Already last year some 32B models maxed it so i think it is sufficient to test
主题标签模型发布端侧推理
原始关键词#benchmark#medical#models#local#test#15b