Thanks for the open source contribution!
One small note is that the diff_spec function differs from common library standards.
- it returns N elements, while numpy/pytorch return N-1.
- it seeds
out[0] with the first value, while pandas uses NaN.
| implementation |
length |
first element |
diff_spec |
N |
a[0] |
| numpy / pytorch |
N-1 |
a[1] - a[0] |
| pandas |
N |
NaN |
Thanks for the open source contribution!
One small note is that the
diff_specfunction differs from common library standards.out[0]with the first value, while pandas uses NaN.diff_speca[0]a[1] - a[0]NaN