How to read the numbers on the output "px_est_0.npy"

[Mujingrui]

Hello Qiao,

Hope all is well!

I am a beginner at estimating densities in the neural network world. I tried your code and ran the example, which was successful. When I input my simulation dataset, the sample size is 250, the dimension is 10, and the result in the file "px_est_0.npy" gives all negative numbers. I assume these numbers are estimated probabilities, so I'm kind of confused on how to read these negative numbers. Thank you so much for your help and time!

Best,

[kimmo1019]

If you set log=True in estimate_py_with_IS, then it is likely you will get negative values as it is the log density. Could you let me know which script or program did you run?

[Mujingrui]

Thank you so much, Qiao!

Sorry, it is my bad. I did not read it carefully. yeah, I set up log = TRUE in estimate_py_with_IS. I trained the model and estimated the density values using estimate_py_with_IS.

May I ask one more question? Since I do not have much GPU resources, I am wondering if it is possible to run the network with CPU in an efficient way? or if it is possible to do some pre-training and refine model later to reduce the running time.

I set up my own hyper-parameters in a .yaml file, I am not sure if it is the key to reduce the running time. Or in your .yaml files, do you pre-train the model and get your hyper-parameters in an initially good sense.

Thank you so much for your help!

Best,