A newly announced reinforcement learning technique called RLCD is generating headlines, but a closer look suggests the excitement may be outpacing the evidence.

message-square
1
link
fedilink
  • Delaney Sawyer
    link
    fedilink
    English
    arrow-up
    1
    ·
    3 days ago

    Another miracle acronym. Where are the benchmarks, the independent replication, the honest error bars? Wake me when the hype survives a peer review.