Closed behrica closed 8 months ago
Yes, there are such plans. I'll try to figure out how to make it properly with minimal effort. The main point of changes is around clustering API (no more double-double arrays input for some functions).
yes, I saw other big changes, but not affecting fastmath I think: https://github.com/scicloj/scicloj.ml.smile/issues/12
But I use indeed the clustering from fastmath: https://github.com/scicloj/scicloj.ml.smile/blob/main/src/scicloj/ml/smile/clustering.clj
Indeed... I was fooled by changes in classification. I'll take a look soon and upgrade it.
Looks like it went smoothly. Some minor API changes in my case:
setSeed
can't be called without argumentsnrows
-> nrow
in Matrix classRoughly tested and all other things worked (clustering), so expect a snapshot soon.
BTW, all the SMILE 3.0.2 deps are:
[com.github.haifengl/smile-base "3.0.2"]
[com.github.haifengl/smile-core "3.0.2"]
[com.github.haifengl/smile-mkl "3.0.2"]
[org.bytedeco/arpack-ng "3.9.0-1.5.9"]
[org.bytedeco/arpack-ng-platform "3.9.0-1.5.9"]
[org.bytedeco/openblas "0.3.23-1.5.9"]
[org.bytedeco/openblas-platform "0.3.23-1.5.9"]
[org.bytedeco/javacpp "1.5.9"]
@behrica I deployed 2.3.0-SNAPSHOT
. If you can test it it would be great. From my perspective clustering
works.
Released 2.3.0 with Smile 3.0.2
I was looking into it for
scicloj.ml.smile
. But as it depends on fastmath, I think it would be better to upgrade fastmath before.