Contents¶ Contents: 📚 Methods Implemented in NeuPI 🔹 Single-Pass Inference & Marginal MAP in Probabilistic Circuits 🔹 Single-Pass Inference & Neural Embeddings for Constrained MPE 🔹 ITSELF Engine for General MPE Inference 🔹 SINE: Enhanced Neural Embeddings and Discretization for Probabilistic Graphical Models Examples NeuPI: Loading and Evaluating PGMs NeuPI: Self-Supervised Training of a Neural Surrogate NeuPI: Inference and Test-Time Refinement NeuPI: Advanced Discretization Methods neupi neupi.core neupi.embedding DiscreteEmbedder ITSELF_Engine ITSELF_Engine.run() IdentityEmbedding KNearestDiscretizer MADE MADE.evaluate() MADE.forward() MADE.update_masks() MLP MLP.forward() MLP.initialize_weights() MarkovNetwork MarkovNetwork.forward() OAUAI SelfSupervisedTrainer SelfSupervisedTrainer.fit() SelfSupervisedTrainer.step() SinglePassInferenceEngine SinglePassInferenceEngine.run() SumProductNetwork SumProductNetwork.evaluate() SumProductNetwork.forward() ThresholdDiscretizer factory() mpe_log_likelihood_loss()