Hi everyone, I was wondering whether anyone knows why t-SNE results into memory errors when reducing the dimension of the data to the default number, which is 2.
I have the mnist dataset and I apply some transformation on it. Resulting in a reduced mnist dataset.
Original: 60000x784
After transformation: 60000x200

When I apply t-SNE on transformed mnist 60000x200, I always get a memory error. Also I was wondering why doesn't the multicore option of n_jobs=-1 exist?