Replies: 1 comment
-
I think it’s mainly because floating point rounding errors often over-shoot or under-shoot the endpoint, so that the strategy you suggest causes the length of the output array to be off by one compared to what the user expects. That said, in the integer case we could probably avoid materialization |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
When
step is None
, alax.iota
is generated. However,jnp.arange(0, 10, 2)
will become a constant. Why isn't it implemented asiota() * step + start
?Beta Was this translation helpful? Give feedback.
All reactions