guyga

4 ms·

1. 1.▲
   by guyga 4y ago

   From the paper, the Polish tests are from May 2022, and the training dataset was from January 2022, so this is unlikely.
2. 2.▲
   Minerva: Solving Quantitative Reasoning Problems with Language Models(ai.googleblog.com)

   98 points by guyga 4y ago|36 comments
3. 3.▲
   BIG-bench: a collaborative language model benchmark from Google and OpenAI(github.com)

   4 points by guyga 6y ago|0 comments