Open
Description
We have 4 forecasters(TCN, S2S, LSTM) in TF2, there definitions are here. https://github.com/intel-analytics/BigDL/tree/main/python/chronos/src/bigdl/chronos/forecaster/tf
Currently, They only support string loss and metric . We want to add the support of
https://www.tensorflow.org/api_docs/python/tf/keras/losses instance for loss and customized metric for metric.
A good reference is #4798 which support customized loss and metric in pytorch forecaster
Checklist:
- Code in each forecaster
- API doc change in each forecaster
- Unittest for customized loss and metric