Anthral Labs retweetledi
We wrote a blog on how we managed to beat some of the frontier labs in forecasting and predicting the future. news.anthral.com/p/how-did-we-r…
also overhauled our website anthral.com
English
Anthral Labs
5 posts
Anthral Labs retweetledi
what do markets of the future look like if insights into the future are democratized
I wouldn't know, but I'd safely bet that the one democratizing the future will make BANK. anthral.com
English
Anthral Labs retweetledi
English
Anthral Labs retweetledi
@anthral_labs has achieved parity with human superforecasters on ForecastBench market questions. This was considered the harder subset of the benchmark with @GoogleDeepMind and @xai models lagging superforecasters (albeit closely)
Forecasting Research Institute@Research_FRI
Does parity on dataset questions mean ForecastBench is “solved”? No. On market questions that require judgment about novel, one-off events, LLMs are still behind. Even on dataset questions, the 64.9% achieved by Google DeepMind is unlikely to be the ceiling of what’s possible. ForecastBench doesn’t stop being useful at human parity. It will continue tracking LLM progress even if they surpass expert human forecasters.
English
Anthral Labs retweetledi
We just ranked 2nd on a forecasting benchmark!
@harsith2002 and I founded @anthral_labs to work on superforcasting. A couple of days ago, we got a rank on forecastbench. 2nd on the market subset and 7th overall
English