The zantetsu-trainer crate is no longer maintained. The ML training infrastructure it contained was removed as part of the zantetsu 0.2 release, which replaced the neural parser with a pure heuristic engine.

A tombstone version (0.2.0) has been published and 0.1.4 has been yanked. There is no replacement for zantetsu-trainer.

If you were using zantetsu-trainer as part of a zantetsu-based filename parsing pipeline, switch to zantetsu 0.2, which requires no training step.