Hutter Prize
A compression competition established in 2006 by AI researcher Marcus Hutter, offering a prize for algorithms that achieve the smallest compressed size of a 100 MB English Wikipedia text corpus. The competition is grounded in Kolmogorov complexity theory and aims to use compression performance as a measure of an AI system's depth of natural language understanding.
Timeline (last 90 days)
Large language models can be interpreted as a type of compressor: the stronger the model's next-token prediction capability, the lower the compression ratio it can theoretically achieve.
The Hutter Prize is a competition that rewards optimal lossless compression of Wikipedia text, built on the hypothesis that better compression is equivalent to better understanding.
The Hutter Prize was established in 2006 by AI researcher Marcus Hutter to reward algorithms that can compress 100MB of English Wikipedia text to the smallest possible size.
All Facts (3)
Large language models can be interpreted as a type of compressor: the stronger the model's next-token prediction capability, the lower the compression ratio it can theoretically achieve.
50%UnverifiedThe Hutter Prize is a competition that rewards optimal lossless compression of Wikipedia text, built on the hypothesis that better compression is equivalent to better understanding.
50%UnverifiedThe Hutter Prize was established in 2006 by AI researcher Marcus Hutter to reward algorithms that can compress 100MB of English Wikipedia text to the smallest possible size.
50%