3 ms·
As usual with ML, I now wonder how “similar” the test set is to the training set, compared to the examples that are neither in the training set, nor the test se
by antipaul 6y ago
As usual with ML, I now wonder how “similar” the test set is to the training set, compared to the examples that are neither in the training set, nor the test set:
TODO: 200 million - 170,000 training - 100 test ~= 199.8 million proteins
- kxs 6y agoThey trained on 170k sequences/ structures/ proteins, each sequence has 10s to 100s or even 1000s amino acids. Structure is much more conserved than sequence. Out of the 100 targets, roughly 1/4th have no similarity to known structures, so there shouldn't be an overlap for those with the training set. They did very well on those targets.