wildjames / predictive_coding_rs / 22832190102

91%

push

github

Generalise the data boundary (#10)

* make the dataset defined by trait, rather than hardcoded. Pass around Arc of implementations, so we have threadsafe, readonly datasets. Also, split the training CPU handlers into separate files

* Add coverage command to makefile

* Some test updates

* Some more tests - coverage still low

* Move common test code to a test_utils file

* Add ETA to the reporting hook

* Add evaluation dataset parameter to the TrainConfig struct

* Add the evaluation dataset to the training configuration - optional field. Update test data.

547 of 559 new or added lines in 12 files covered. (97.85%)

1506 of 1643 relevant lines covered (91.66%)

16.57 hits per line