跳到正文
Hacker News:AI 熱帖·· 14 天前AI 評分17

MaxxBench:DeepSWE 的高分,真的反映實力嗎?

MaxxBench – "Nice DeepSWE score. But is it true?"

AI 導讀

MaxxBench 質疑 DeepSWE 的理想評分是否反映真實能力。Hacker News 討論者指出,基準測試與真實編程能力可能存在差距,亦有人分享使用 DeepSeek 的負面體驗;另有留言提到,該基準由兩兄弟建立。

正文

Image 1Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

MaxxBench – "Nice DeepSWE score. But is it true?" (unlap.ai) 26 points by antiuniverse1 hour ago | hide | past | favorite | 4 comments help Image 2Neywiny17 minutes ago | next[–] This is perfect. Real programming has a ton of these instances of minor changes. reply Image 3saejox8 minutes ago | prev | next[–] There is always a huge discrepancy between benchmarks and real world capabilities. For my use case deepseek 4.1 flash is almost as dumb as haiku 5.5. Both are fast garbage producers. reply Image 4jeffbee2 minutes ago | parent | next[–] I was beginning to worry I had overlooked the inner beauty of DeepSeek. When I have tried it, it was totally useless. Never understood HN cheering for it. reply Image 5brcmthrowaway19 minutes ago | prev[–] Two brothers created a benchmark, family is so proud. reply Image 6 Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact Search:

來源:Hacker News:AI 熱帖 · news.ycombinator.com