johnma2006

3 ms·

1. 1.▲
   by johnma2006 3y ago

   The key here is that the query sample is uniformly distributed, and that this is sufficient. I think some other comments in this thread give some decent intuitions why this is true. Cheers
2. 2.▲
   by johnma2006 3y ago

   Their estimator does not require any assumption of uniformity. Notice that P(random ID is valid) = total_valid_IDs / total_IDs is the same regardless of the distribution of valid IDs.
3. 3.▲
   Implementation of Mamba in one file of PyTorch(github.com)

   417 points by johnma2006 3y ago|108 comments