Top all-time · #12504
Lennart Heim
Personal blog · blog.heim.xyzaipolitics
Lennart Heim's personal blog at blog.heim.xyz covers ai and politics. Since 2022, 6 different Hacker News readers have linked it 6 times, most often “Estimating PaLM's Training Cost”. It is still writing: the last post was on 2025-04-29, and you can follow it by feed.
Too few sampled comments (2) to say how often HN recommends it.?
Why HN links it
HN readers link Lennart Heim's posts on AI training compute: the estimated cost of training Google's PaLM model, and the FLOP versus FLOPS unit terminology. Commenters cite them as sources.
Summarised by Claude from a sample of 2 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Musings on mostly AI Governance, by Lennart Heim.
Most-linked posts?
- Estimating PaLM's Training Cost3 people · 3 mentions · first linked 2022-04-07
What Hacker News says?
Pet peeve how they write FLOPS in the figure when they meant FLOP. Maybe the plural s after FLOP got capitalized. https://blog.heim.xyz/flop-for-quantity-flop-s-for-performan...
Comment linking /flop-for-quantity-flop-s-for-performance/amp/
on “AI 2027”Submitted Estimating PaLM's Training Cost as a story · 51 points
HN discussionSeems odd to link the tweet, but not the article directly linked to the tweet. If you're interested, but don't want to click through, the actual blog post is below. https://blog.heim.xyz/palm-training-cost/
Comment linking /palm-training-cost/
on “It costs you $9M to $17M to train PaLM (Google's 530B param. language model)”Submitted Google's 540B language model (PaLM) training costs $9M to $17M as a story · 5 points
HN discussionSubmitted China will likely match U.S. AI model capabilities this year as a story · 3 points
HN discussionSubmitted The Rise of DeepSeek: What the Headlines Miss as a story · 1 points
HN discussion
