claude new model funny bullshit
講乜都話trigger safety measure 然後唔比繼續
等睇神prompt了

法國D乙女手機驚好鬼harmsub 祖國個D harmsub手機驚做唔做到?
https://www.youtube.com/watch?v=uyC7FHhcyGU

法國D乙女手機驚好鬼harmsub 祖國個D harmsub手機驚做唔做到?
https://www.youtube.com/watch?v=uyC7FHhcyGU

呢隻不就是祖國harmsub手機驚?

法國D乙女手機驚好鬼harmsub 祖國個D harmsub手機驚做唔做到?
https://www.youtube.com/watch?v=uyC7FHhcyGU

呢隻不就是祖國harmsub手機驚?

男性向個D有冇咁harmsub?

we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms. Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).

wtf?

we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms. Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).

wtf?

簡單講 佢覺得你問既問題關於machine learning
佢可以勁up個錯既答案 或者扮唔識 唔話你知

we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms. Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).

wtf?

簡單講 佢覺得你問既問題關於machine learning
佢可以勁up個錯既答案 或者扮唔識 唔話你知

we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms. Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).

wtf?

簡單講 佢覺得你問既問題關於machine learning
佢可以勁up個錯既答案 或者扮唔識 唔話你知

唔比你用LLM整LLM 好像有點道理?

轉會用composer好過啦
claude is shit now

轉會用composer好過啦
claude is shit now

咁又唔係 有人話同opus差唔多價
因為fable醒太多 用比較少既token已經解決到問題
不過太多safe guard搞到變joke 但比如做reverse engineering幾勁下

當然唔會話reverse engineering
話係debug/呢個project既新dev想了解下呢個project/買左個software用唔到 廠商唔回應想"fix"左佢 等等

轉會用composer好過啦
claude is shit now

咁又唔係 有人話同opus差唔多價
因為fable醒太多 用比較少既token已經解決到問題
不過太多safe guard搞到變joke 但比如做reverse engineering幾勁下

當然唔會話reverse engineering
話係debug/呢個project既新dev想了解下呢個project/買左個software用唔到 廠商唔回應想"fix"左佢 等等

點workaround
any jailbreak?

轉會用composer好過啦
claude is shit now

咁又唔係 有人話同opus差唔多價
因為fable醒太多 用比較少既token已經解決到問題
不過太多safe guard搞到變joke 但比如做reverse engineering幾勁下

當然唔會話reverse engineering
話係debug/呢個project既新dev想了解下呢個project/買左個software用唔到 廠商唔回應想"fix"左佢 等等

點workaround
any jailbreak?

上面講左

轉會用composer好過啦
claude is shit now

咁又唔係 有人話同opus差唔多價
因為fable醒太多 用比較少既token已經解決到問題
不過太多safe guard搞到變joke 但比如做reverse engineering幾勁下

當然唔會話reverse engineering
話係debug/呢個project既新dev想了解下呢個project/買左個software用唔到 廠商唔回應想"fix"左佢 等等

點workaround
any jailbreak?

上面講左

ok 姐係暫時只可以係跟需要講大話
無一個general既方法workaround

轉會用composer好過啦
claude is shit now

咁又唔係 有人話同opus差唔多價
因為fable醒太多 用比較少既token已經解決到問題
不過太多safe guard搞到變joke 但比如做reverse engineering幾勁下

當然唔會話reverse engineering
話係debug/呢個project既新dev想了解下呢個project/買左個software用唔到 廠商唔回應想"fix"左佢 等等

點workaround
any jailbreak?

上面講左

ok 姐係暫時只可以係跟需要講大話
無一個general既方法workaround

不嬲冇 有咁易既話自然會patch左
講到尾只係blacklist 要唔trigger佢D keyword或者scenario

轉會用composer好過啦
claude is shit now

咁又唔係 有人話同opus差唔多價
因為fable醒太多 用比較少既token已經解決到問題
不過太多safe guard搞到變joke 但比如做reverse engineering幾勁下

當然唔會話reverse engineering
話係debug/呢個project既新dev想了解下呢個project/買左個software用唔到 廠商唔回應想"fix"左佢 等等

點workaround
any jailbreak?

上面講左

ok 姐係暫時只可以係跟需要講大話
無一個general既方法workaround

不嬲冇 有咁易既話自然會patch左
講到尾只係blacklist 要唔trigger佢D keyword或者scenario

好 以為可以好似d 19 chatbot咁中間隔下字 轉下碼黎workaround!
btw cohere gd?好似多人講

kid又有新rune
on lio