19
Had enough of all those dragged out videos, so I used a LM to defeat them. (2 Minutes) (www.youtube.com)
submitted 2 years ago by to c/videos@lemmy.world
6 comments fedilink hide all child comments
 

You know how sometimes you see an intriguing title and just want to know the answer, only for the youtuber to drag it out for 5 minutes until he actual answers the thing.

Not anymore with this prompt, I used it so much that I end up being less on youtube. This method will optimize 10 minutes of inefficiently placed information down to its core of a few sentences.

you are viewing a single comment's thread
view the rest of the comments
[–] 20 points 2 years ago (3 children)

Yeah, I'm not watching this two minute video. Humans of the Internet, what is the prompt?

  • permalink
  • fedilink
  • source
  • hideshow 6 child comments
    •
  • [–] [S] 9 points 2 years ago* (last edited 2 years ago)

    lmao, based. The video is designed to show the method to people who never used a LM before in their life, just in case. The prompt is (its also in the description of the video):

    Please read the following transcript of a youtube video and answer this: "(insert title between these quotation marks)"

This is the transcript of the video:
(insert transcript here)

    make sure to turn off timestamps in the transcript before copying it.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 3 points 2 years ago (1 child)

    Watched it

    It's not a specific prompt, it's about how to get the transcript through the UI so you can paste it in for the summary. Most of the video is about other things

    I personally prefer using the highlights in SponsorBlock (which to be fair this person recommendeds as well)

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] [S] 1 point 2 years ago (1 child)

    I'd just like to note that you should indeed use the specific prompt in the description, or in my other reply here, to get a proper response from the LM. The prompt is designed to have the LM answer based only on the transcript and with a hard focus on the title, that made you click the video in the first place. That way your curiosity is sated and you can move on without having to deal with 10 minutes of bloat.

    It takes the power of an LM being able to process language and the information assembled by a human. That way the LM cant drift into ambiguity or guessing.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •