alexlyzhov

• alexlyzhov 16 Feb 2023 18:07 UTC

  5 points

  2 ∶ 0

  on: Straight­for­wardly elic­it­ing prob­a­bil­ities from GPT-3

  I tried the Anthropic model on this dataset with roughly your prompt and it’s much better in terms of KL divergence between its predictions and Manifold probabilities. Giving it 10 web search results in a prompt further improves the performance. But the difference no search → search is smaller compared to GPT-3 → Anthropic, I’d say mainly because of unhelpful search results.

• alexlyzhov 17 Jul 2022 7:08 UTC

  1 point

  0 ∶ 0

  on: New Data Vi­su­al­i­sa­tions of the EA Forum

  Nice visualization, I wonder how it’s doing now? Curious about any post-mortem comments.

• alexlyzhov 19 May 2022 0:42 UTC

  13 points

  0 ∶ 0

  in reply to: Nick Whitaker’s comment on: Bad Omens in Cur­rent Com­mu­nity Building

  I think it’s not free books per se, but free books related to phrases “here’s what’s really important”, “this is how to think about morality” that are problematic in the context of the Bible comparison

• alexlyzhov 19 Aug 2021 16:00 UTC

  4 points

  0 ∶ 0

  in reply to: Nathan Young’s comment on: What EA pro­jects could grow to be­come megapro­jects, even­tu­ally spend­ing $100m per year?

  Some people have been saying that Starlink system’s limit is 0.5M consumers even after they release a whole lot more satellites: https://​​www.techdirt.com/​​articles/​​20200928/​​09175145397/​​report-notes-musks-starlink-wont-have-capacity-to-truly-disrupt-us-telecom.shtml. This would mean you can’t expect it to turn even 0.1% of unconnected people to netizens.