`pyro.infer.Trace_ELBO` is a class in the Pyro library for variational inference in probabilistic programming using the Evidence Lower Bound (ELBO) loss function. It computes an approximation of the marginal likelihood by constructing a trace of the model's execution and calculating the sum of the log probabilities of all encountered latent variables. This class is commonly used as an objective function to optimize the parameters of a model in order to fit the observed data.
Python Trace_ELBO - 30 examples found. These are the top rated real world Python examples of pyro.infer.Trace_ELBO extracted from open source projects. You can rate examples to help us improve the quality of examples.