Filter items, where token splits of the wrong and the correct identifiers are equal (they differ in non-letter symbols or case spelling).
Filter items, where wrong and correct identifiers are equal on lemmas level.
The result is obtained with the updated TokenParser, which is not included in the last src-d/ml release. Therefore, to have results reproducible by default, we either need to make a release, or load the package from the github on setup.
The result is obtained with the updated TokenParser, which is not included in the last src-d/ml release. Therefore, to have results reproducible by default, we either need to make a release, or load the package from the github on setup.