Arnaldo Leon

I ran into the same issue, and the work-around is to subclass the Tuner class and re-implement `run_trial`, so you can bypass the checkpoint mechanism. ```python class BayesianTuner(keras_tuner.BayesianOptimization): def __init__(self,...

This is a usage error, not a linking error. It seems that your `ldb` parameter is not correct. From the [documentation](https://www.netlib.org/lapack/explore-html/d1/d54/group__double__blas__level3_gaeda3cbd99c8fb834a60a6412878226e1.html): > On entry, LDB specifies the first dimension of...

I just ran into this same problem just now. This is also happening on Linux. It seems you cant build a static library with iomp5. So for example: ```toml [dependencies]...

Ok I found a work-around: the `libmkl_core.a` library is in a different directory than `libiomp5.a`. So what I did is just create a symlink: ``` $ ls libiomp5.a -> ../../../compiler/lib/intel64_lin/libiomp5.a...

I've pushed a potential fix https://github.com/rust-math/intel-mkl-src/pull/63