August 26, 2023@elonmusk →
“@mezaoptimizer Not that it’s much consolation, but the energy efficiency of LLMs using auto-regressive transformers is extremely bad, not just in training, but also inference. I think it’s off by several orders of magnitude.”
Engagement vs. median curated tweet
874
Likes
46
Retweets
81
Replies
Other tweets from the same period
“@X___PB @TimSweeneyEpic I’m guessing that the single largest use of that power is making the cursor …”
Aug 29, 20231.2K likes
“@TimSweeneyEpic But let’s say you had a choice between large size or small, which would you pick?”
Aug 29, 2023852 likes
“@INTERIORPORN1 Throw some sounds of groaning metal support beams failing to spice it up”
Aug 29, 20232.0K likes
“@karpathy Elden Ring is incredible. I liked Cyberpunk. Baldur’s Gate 3 is supposed to be great.”
Aug 29, 20231.9K likes
