I am Using tensor flow probability to build time series forecasting through BSTS

TensorFlow General Discussion
,
1

But every time I rerun my model forecasting numbers are changing . I don’t know how to use Random seed here. Can anyone please help me with that. I am stuck here from last two days.

3

Hi @VEERESH_G ,

To get consistent model forecasts, set a random seed to control randomness in libraries like TensorFlow . For TensorFlow/Keras, set seeds for TensorFlow, NumPy, and Python’s random module. Optionally, disable GPU nondeterminism and set seeds for layers like Dropout .

Thank You .