• 1 Post
  • 65 Comments
Joined 1 year ago
cake
Cake day: July 3rd, 2023

help-circle


  • I don’t like the idea of restricting the model’s corpus further. Rather, I think it would be good if it used a bigger corpus, but added the date of origin for each element as further context.

    Separately, I think it could be good to train another LLM to recognize biases in various content, and then use that to add further context for the main LLM when it ingests that content. I’m not sure how to avoid bias in that second LLM, though. Maybe complete lack of bias is an unattainable ideal that you can only approach without ever reaching it.












  • How much of the music at that EDM festival you went to will you listen to again? Will you buy any of their records? Or maybe just add a few of their tracks to a playlist on Spotify such that you may hear that music a couple more times in the future?

    I may be making assumptions about you, but I think Rick’s point about new music becoming “disposable” because there’s just so much of it coming out remains valid. There’s still new songs coming out that transcend this, but the musicians making this music are finding it harder and harder to build up a following, get exposure, and convince record labels to invest in them, when these labels can invest way less, produce a shitton of disposable music and get a better (short-term) ROI.