The results of using FP model for MMNIST are really bad

[gitzyong812]

My results of using fix-prior model for MMNIST generation are really bad. As the following gif shows, after 25 epochs. I have tried the code with over 100 epochs, but the results don't improve much.

I don't find the comparison of FP and LP over MMNIST in the paper. What's your results in such case? Anybody get good results using FP on MMNIST? Thanks!

[xjgaocs]

I got a similiar results as yours using fp setting. Actually, I haven't got a better results of fp than lp.

[DanielTakeshi]

SVG-LP seems to be working well, maybe try that over FP.

From the command in the README, I have this after 25 epochs:

Here's the result after 300 epochs, i.e., this is the sample_299.gif produced from the code: