Canadian actress.

V1.0 Best to use with pair embedding (HERE), in such case adjust weights as necessary.

V2.0 Standalone version with no pair embedding, result of an experimental training method. Any feedback welcomed.