Tags: mlverse/tabnet
Tags
multi-outcome `tabnet-fit` and `predict` (#118) * add multi-output and multilabel test * switch to `expect_no_error()` for readability * `momentum` consistently default to 0.02 switch target `y` from vector to array * turn `output_dim` into vector when multi_output manage loss cases for multi_output lint and refactor code * pass `is_multi_outcome` to predict encode output_dim for multi-outcome improve multi-outcome classification loss split predict based on `is_multi_outcome` * working predict_impl_class and predict_numeric switch to hardhat v1.3.0 * refactor predict_impl_ for a clearer case_when() call * improve `check_type` to manage multi-outcome fix tests vqlues add mixed-outcome and multi-outcome with valid test * add consistency checks for outcome types * add multi-output description in `tabnet-fit` move multi-output tests is a dedicated file * improve multi-outcome tests fix multi-outcome classification