jellyberg

5 ms·

1.
by jellyberg 3y ago
Forecasting tournaments: https://www.metaculus.com/ https://www.infer-pub.com/ Forecasting your own life: https://fatebook.io/ Training your calibration: https://programs.clearer
2.
by jellyberg 3y ago
See https://samotsvety.org/track-record/
3.
by jellyberg 3y ago
Ah thank you, I missed this!
4.
by jellyberg 3y ago
There are four examples in the first category (imagination), there are five more categories with lots of other examples. Currently, there's no API for Gemini Pro, so you need to use bard.google.com to ask it questions.
5.
Compare how Gemini Pro and GPT 3.5 answer the same questions(theaidigest.org)
2 points by jellyberg 3y ago|4 comments
6.
by jellyberg 3y ago
Yeah - worth noting that we use temperature=0 for reproducibility while ChatGPT I think uses t=0.7. We also prefix the prompt with few-shot examples of questions and answers with chain of thought examples to elicit the models' full cap
7.
by jellyberg 3y ago
Yeah it's hard to compare across models, interested in suggestions here. We give all models a bunch of few-shot examples, which improves GPT-3 (davinci)'s question answering substantially. GPT-2 sometimes generates something that
8.
Compare how GPT-2, 3, 3.5 and 4 answer the same questions(theaidigest.org)
82 points by jellyberg 3y ago|33 comments
9.
Show HN: Fatebook – the fastest way to make and track predictions(fatebook.io)
4 points by jellyberg 3y ago|0 comments
10.
by jellyberg 3y ago
Some examples of different kinds of resolution criteria for forecasting questions: Personal mechanisms “If we choose this HR provider, will I think it was a good idea in two month’s time?” “Each day I’ll write down whether I want to leave o
11.
Show HN: Fatebook – Superforecasting in your Slack(fatebook.io)
5 points by jellyberg 3y ago|1 comments