News for hackers. Stuff that matters.
Re:DeepSeek-R1 by dcreater on Monday January 20, 2025 @07:12PM (#42774742)
Why ask it for a joke? That's such a bad way to try out a reasoning model
Re:DeepSeek-R1 by HeatrayEnjoyer on Monday January 20, 2025 @07:21PM (#42774801)
It's a great test. Creating a joke that is both funny and novel requires many steps.
Re:DeepSeek-R1 by Weetile on Monday January 20, 2025 @07:46PM (#42775004)
Creating a joke is a great demonstration of a multistage reasoning/deduction process
Re:DeepSeek-R1 by simonw on Monday January 20, 2025 @07:21PM (#42774798)
Watching them attempt to reason their way to a joke is really amusing.
Re:DeepSeek-R1 by qqqult on Tuesday January 21, 2025 @01:49AM (#42777274)
I asked deepseek-14b for a joke in the style of Norm Mcdonald, and it delivered an overexplained deadpan absurd joke that I thought was pretty similar to Norms style. Then I tried again with different styles and realized that all jokes it tries to make are in that same style as if it tries to follow a manual on how to construct a joke without actually being funny. Which is pretty funny
Re:DeepSeek-R1 by marci on Tuesday January 21, 2025 @05:59AM (#42778725)
Would that be funny? Failing a job interview because, when ask to make a joke about a pelican and a walrus who run a tea room together, you couldn't come up with a quip, therefore showing your inability to reason.

Let's not even talk about the "r" you forgot when asked to write "cranberry"...