Replies: 1 comment 1 reply
|
Dear @shuailiu199966, thank you for reporting this issue. You are totally correct, we did change the interface as you described. The reason for this is that I wanted the first arguments of both random_state and zero_state to be identical. The random_state function previously did not have the option to generate multiple columns, this was a new feature. In general, I try to avoid any changes in the interface, but in this case the change was necessary. Thank you also for pointing out that the documentation was not updated, I missed that. I now updated the documentation. Let me know if everything checks out now. All the best, Alex |
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Dear Author,
After upgrading to the new version 0.4.1, I observed that the behavior of the random_state function appears to have changed from previous versions. According to the documentation, the function is declared as:
State random_state(Block const &block, bool real = true, int64_t seed = 42, bool normalized = true);
Here, the parameter int64_t seed = 42 is documented as the random seed. However, in version 0.4.1, this parameter no longer seems to function as the random seed. Instead, it appears to control the number of columns in the generated state.
For example, when I passes a block with a dimension of 64 and sets seed to 10, the function returns a State that is a 64×10 matrix, rather than a random state generated using the seed value of 42. This behavior is inconsistent with how the parameter functioned as a random seed in earlier versions.
I would like to confirm whether this change is intentional or if it might be a bug introduced in the new version. Thank you for your attention and clarification.
All reactions