lxuechen / bdmc

PyTorch implementation of Bidirectional Monte Carlo, Annealed Importance Sampling, and Hamiltonian Monte Carlo.

I like big neural networks and good datasets.

bdmc's Issues

Error

Traceback (most recent call last):
File "bdmc.py", line 67, in
main()
File "bdmc.py", line 63, in main
bdmc(model, loader, forward_schedule=np.linspace(0., 1., 500), n_sample=100)
File "bdmc.py", line 33, in bdmc
forward_logws = ais_trajectory(model, load, mode='forward', schedule=forward_schedule, n_sample=n_sample)
File "../BDMC/ais.py", line 76, in ais_trajectory
log_int_1 = log_f_i(current_z, batch, t0)
File "../BDMC/ais.py", line 42, in log_f_i
log_likelihood = discretized_logistic(*model.decode(z), data)
TypeError: discretized_logistic() takes 3 positional arguments but 10001 were given

I got this error simply running your code. How can I fix it?