`metrics` also could be any python object. If users use NNI built-in tuner/assessor, `metrics` follows the same format rule as that in `report_intermediate_result`, the number indicates the model's performance, for example, the model's accuracy, loss etc. This `metrics` is reported to [tuner](tuners.md).
`metrics` also could be any python object. If users use NNI built-in tuner/assessor, `metrics` follows the same format rule as that in `report_intermediate_result`, the number indicates the model's performance, for example, the model's accuracy, loss etc. This `metrics` is reported to [tuner](Builtin_Tuner.md).