Code for our paper Delta-AI: Local objectives for amortized inference in sparse graphical models by Jean-Pierre Falet, Hae Beom Lee, Nikolay Malkin, Chen Sun, Dragos Secrieru, Dinghuai Zhang, Guillaume Lajoie, and Yoshua Bengio.
Synthetic tasks
python synthetic/main_delta.py --alg rand --model ising --graph lattice --vdim 264 --epsilon 1 --temp 10 --glr 1e-3 --mlr 1e-1
Latent-variable modeling (MNIST)
python mnist/main_delta.py --alg rand --sampling_dag partial --epsilon 0.05 --temp 4 --q_lr 1e-3 --p_lr 1e-3 --marg_q_lr 1e-1 --marg_p_lr 1e-1 --q_objective delta