# StackOverflowError in Bayesian Neural Networks Tutorial
**URL:** https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372
**Category:** Machine Learning
**Tags:** question
**Created:** [March 23, 2020, 2:05am UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372 "2020-03-23T02:05:38Z")
**Posts on this page:** 6
**Page:** 1
### Author:  [@Farlein](https://discourse.julialang.org/u/Farlein)
#### Post date: [March 23, 2020, 2:05am UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372/1 "2020-03-23T02:05:38Z")
Hi there,
I am learning Bayesian Neural Networks (BNN) using Turing. I have copied the codes from the tutorial, [https://turing.ml/dev/tutorials/3-bayesnn/](https://turing.ml/dev/tutorials/3-bayesnn/).
The original code trains a BNN model with a synthetic dataset with 80 rows. The step “ch = sample(bayes\_nn(hcat(xs…), ts), HMC(0.05, 4), N);” costs 0:02:03 on my machine. If I change the “N = 80” to N=800, it costs 0:03:25. Pretty fast! However, if I change it N=8000, it gives me the error “StackOverflowError”. I have copied some rows of the detailed error information at the bottom of this post.
I want to build a BNN model to predict Admission Yield, and the dataset has about 40,000 rows and 90 variables, so I need to learn how to train a BNN model with relatively large dataset. Would you please help me to solve the error? Please let me know if I need to provide any other information.
Thanks,
Chuan
**StackOverflowError:**
in top-level scope at Learn Turing\_20200316.jl:105
in sample at Turing\azHIm\src\inference\Inference.jl:136
in #sample#1 at Turing\azHIm\src\inference\Inference.jl:136
in sample at Turing\azHIm\src\inference\Inference.jl:148
in #sample#2 at Turing\azHIm\src\inference\Inference.jl:149
in Sampler at Turing\azHIm\src\inference\hmc.jl:302
in DynamicPPL.Sampler at Turing\azHIm\src\inference\hmc.jl:310
in Turing.Inference.HMCState at Turing\azHIm\src\inference\hmc.jl:533
in #HMCState#52 at Turing\azHIm\src\inference\hmc.jl:562
in sample\_init at AdvancedHMC\haUrH\src\sampler.jl:13
in phasepoint at AdvancedHMC\haUrH\src\hamiltonian.jl:129
in phasepoint at AdvancedHMC\haUrH\src\hamiltonian.jl:59
in ∂H∂θ at AdvancedHMC\haUrH\src\hamiltonian.jl:28
in ∂logπ∂θ at Turing\azHIm\src\inference\hmc.jl:401
in gradient\_logp at Turing\azHIm\src\core\ad.jl:73
in gradient\_logp\_reverse at Turing\azHIm\src\core\ad.jl:141
in at Tracker\cpxco\src\back.jl:149
in #18 at Tracker\cpxco\src\back.jl:140
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at base\abstractarray.jl:1921
in #16 at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
in foreach at Tracker\cpxco\src\back.jl:113
in back at Tracker\cpxco\src\back.jl:125
in back\_ at Tracker\cpxco\src\back.jl:113
---
### Author:  [@mohamed82008](https://discourse.julialang.org/u/mohamed82008)
#### Post date: [March 23, 2020, 2:33am UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372/2 "2020-03-23T02:33:57Z")
I have seen this error before and it seems to be a Tracker issue with large loops. Zygote doesn’t have this problem. If you go on Turing#master you can use Zygote for AD with:
```julia
using Zygote, Turing; Turing.setadbackend(:zygote)
```
However, Zygote will take a lot of memory when compiling the gradient the first time.
---
### Author:  [@Farlein](https://discourse.julialang.org/u/Farlein)
#### Post date: [March 23, 2020, 1:28pm UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372/3 "2020-03-23T13:28:07Z")
Thanks, Mohamed. I have added Turing#master and am testing it with Turing.setadbackend(:zygote). It runs!
However, for 8000 rows, it is estimated to cost 10:39:00, which is much longer than 0:03:25 for N=800.
Thanks,
Chuan
---
### Author:  [@Farlein](https://discourse.julialang.org/u/Farlein)
#### Post date: [March 23, 2020, 1:35pm UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372/4 "2020-03-23T13:35:44Z")
Another question I want to ask is about ForwardDiff. If I use Turing.setadbackend(:forward\_diff), the program runs fast with 8000 rows for 0:05:40. However, the acceptance rate is constantly 0 for the 5000 samples by HMC, and thus the std is just 0 for nn\_params. It is not the case with 80 rows. With 80 rows, I see the std is larger than 0 for each nn\_params in the chain ch from “ch = sample(bayes\_nn(hcat(xs…), ts), HMC(0.05, 4), N);”.
---
### Author:  [@mohamed82008](https://discourse.julialang.org/u/mohamed82008)
#### Post date: [March 24, 2020, 7:49am UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372/5 "2020-03-24T07:49:29Z")
With HMC, as you increase the number of data points, you need to lower the step size. Otherwise, just use NUTS.
---
### Author:  [@Farlein](https://discourse.julialang.org/u/Farlein)
#### Post date: [March 24, 2020, 12:35pm UTC](https://discourse.julialang.org/t/stackoverflowerror-in-bayesian-neural-networks-tutorial/36372/6 "2020-03-24T12:35:41Z")
Thanks for your reply, Mohamed. I will lower the step size.