Missed TensorFlow Dev Summit? Check out the video playlist. Watch recordings

tfp.experimental.substrates.numpy.mcmc.CheckpointableStatesAndTrace

View source on GitHub

States and auxiliary trace of an MCMC chain.

@staticmethod
tfp.experimental.substrates.numpy.mcmc.CheckpointableStatesAndTrace(
    _cls, all_states, trace, final_kernel_results
)

The first dimension of all the Tensors in the all_states and trace attributes is the same and represents the chain length.

Attributes:

  • all_states: A Tensor or a nested collection of Tensors representing the MCMC chain state.
  • trace: A Tensor or a nested collection of Tensors representing the auxiliary values traced alongside the chain.
  • final_kernel_results: A Tensor or a nested collection of Tensors representing the final value of the auxiliary state of the TransitionKernel that generated this chain.