IBM / unitxt / 15421788639

81%

push

github

Refactor JaccardIndex and StringContainment metrics to use ReductionInstanceMetric (#1816)

* Refactor JaccardIndex and StringContainment metrics to use ReductionInstanceMetric

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Add JaccardIndexWords metric and corresponding tests

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Update StringContainment metric to use Any type for predictions and references

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Update spearmanr and mean squared error

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Return old metric and fix old tests

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Update documentation and test of JaccardIndex

Renamed JacardIndexWords to JaccardIndexString and made it more general by default (not providing the split).

Also moved to regex tokenzier in jacard_index_words to make it work with multiple spaces.

Signed-off-by: Yoav Katz <katz@il.ibm.com>

* Added documentation and checking for MeanSquaredError

Signed-off-by: Yoav Katz <katz@il.ibm.com>

* Updated json

Signed-off-by: Yoav Katz <katz@il.ibm.com>

* Fix spearmanr

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Added RMSE metric

Signed-off-by: Yoav Katz <katz@il.ibm.com>

* Format

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Revert naming

Signed-off-by: elronbandel <elronbandel@gmail.com>

* Add spearmanr_p_value to evaluation metrics in TestAPI

Signed-off-by: elronbandel <elronbandel@gmail.com>

* FIx tests

Signed-off-by: elronbandel <elronbandel@gmail.com>

---------

Signed-off-by: elronbandel <elronbandel@gmail.com>
Signed-off-by: Yoav Katz <katz@il.ibm.com>
Co-authored-by: Yoav Katz <katz@il.ibm.com>
Co-authored-by: Yoav Katz <68273864+yoavkatz@users.noreply.github.com>

1687 of 2069 branches covered (81.54%)

Branch coverage included in aggregate %.

10432 of 12985 relevant lines covered (80.34%)

0.8 hits per line