Posterior sample problem

[SaxonNB]

Hi, all
I'm meeting a problem that i try to run NPE to calibration a simulator which generate time series data.
But when i follow the instructions on the website. he trocess of train can successfully run, while when i sample from the posterior, the sampling process may get stuck.
Here is my code:
pt_file_path = "./dataset/pt/train_trimmed.pt"
theta, x = process_and_convert_to_tensors(pt_file_path)
x_o = torch.tensor(mean_prices.values)

from sbi.inference import NPE_A

inference = NPE_A(prior=prior)
proposal = prior

_ = inference.append_simulations(theta, x, proposal=proposal).train()
posterior = inference.build_posterior().set_default_x(x_o)
proposal = posterior

posterior_samples = posterior.sample((1,))

[mateuszlickindorf]

Hey,
Although on a different model, I think we might be experiencing the same issue. It could be related to the dimensionality of your model, but more specifics on what exactly you're trying to optimise are needed. Here is the link to my issue where I got a helpful answer:

#1314 (reply in thread)

Hope this helps in any way!
Mateusz