README.md
1# TensorFlow Lite model coverage tests
2
3Various conversion tests on popular mobile models.
4
5
6## Golden values
7
8Some tests rely on pre-computed golden values. The main goal is to detect
9changes affecting unintended parts of TFLite.
10
11Should a golden value test fail after an intended change, the golden values can
12be updated with the following command:
13
14```
15bazel run //third_party/tensorflow/lite/testing/model_coverage:<target> --test_output=all -- --update_goldens
16```
17
18Notice `bazel run` instead of `bazel test` and the addition of the
19`--update_golden` flag.
20
21The updated golden data files must then be included in the change list.
22