Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

turing.guru

Post reply on HN
Print view View original

1–10 of 15 posts

Re: Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

#3

Post by coolio222 »

Does anyone knows the difference to other languages like STAN?

Both use Hamiltonian Monte Carlo. As far as I know STAN cannot model factors but can restrict a variable to integers. In this case, Turing.jl and STAN do the same and switch back to slower MCMC.

So why should I switch to turing.jl?

Their wiki shows that turing.jl is about 10 times slower than STAN:

  Results
  Model 	Turing 	Stan 	Ratio
  Simple Gaussian Model 	1.2 	0.06 	20.24
  Bernoulli Model 	1.53 	0.05 	32.73
  School 8 	2.34 	0.1 	24.41
  Binormal (sampling from the prior) 	0.88 	0.11 	8.37
  Kid 	32.66 	4.72 	6.92
  LDA-vec 	23.94 	3.78 	6.34
  LDA 	72.78 	3.78 	19.28
  Mixture-of-Categorical 	22.28 	6.41 	3.48
  NOTE 1
  Numbers here are inference time in second - smaller number   indicates better performance.

Re: Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

#5

Post by StefanKarpinski »

peatmoss wrote: post #4

Am I missing something here, or is there zero content about this project linked from the landing page?

I think the wiki is the documentation: https://github.com/yebai/Turing.jl/wiki. It's linked in the upper right corner. I agree that the fact that the three big items on the page are not links is a bit confusing.

Re: Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

#6

Post by ChrisRackauckas »

coolio222 wrote: post #3

Does anyone knows the difference to other languages like STAN? Both use Hamiltonian Monte Carlo. As far as I know STAN cannot model factors but can restrict a variable to integers. In this case, Turing.jl and STAN do the same and switch back to slower MCMC. So why should I switch to turing.jl? Their wiki shows that turing.jl is about 10 times slower than STAN: Results Model Turing Stan Ratio Simple Gaussian Model 1.2…

We're setting up parameter estimation in DifferentialEquations.jl with both Stan.jl and Turing.jl as separate but connected projects. For generated data from the Lotka-Volterra equations, Turing.jl seems to be able to recover the four parameters well in minutes, whereas Stan was running for a few hours and didn't get it right. It's hard to know if it's because of the difference in the ODE solver and the accuracy there (Stan has to use its internal solvers, whereas Turing.jl gets to use our whole stack. For testing we chose Tsit5 which should be to some extent equivalent to their rk45, but without knowing the details of their solver it's hard to know how different it truly is) or due to the chosen method for MCMC method. One major advantage of Turing.jl though is it lets us use our full stack: Distributions.jl, all of JuliaDiffEq (ODE/SDE/DAE/DDE/etc. solvers), whereas Stan requires you use their two ODE solvers. So we're going to finish up this DiffEq->Stan automatic generation bridge for testing and benchmarking purposes, but the Turing.jl route looks much more promising. They have done a really good job.

Re: Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

#7

Post by peatmoss »

StefanKarpinski wrote: post #5
peatmoss wrote: post #4

Am I missing something here, or is there zero content about this project linked from the landing page?

I think the wiki is the documentation: https://github.com/yebai/Turing.jl/wiki . It's linked in the upper right corner. I agree that the fact that the three big items on the page are not links is a bit confusing.

Ah, thank you. Apparently dividing my screen vertically made my web browser too narrow, and thus the links disappeared entirely.

Re: Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

#8

Post by bluetwo »

peatmoss wrote: post #4

Am I missing something here, or is there zero content about this project linked from the landing page?

It really does need a better explanation right on the page. It is ridiculous to link here and require clicking on a hard to find link to get actual information.

Poor communication.

Re: Turing.jl: A Fresh Approach to Probabilistic Programming in Julia

#9

Post by cristoperb »

peatmoss wrote: post #7
StefanKarpinski wrote: post #5

Earlier quoted context omitted.

I think the wiki is the documentation: https://github.com/yebai/Turing.jl/wiki . It's linked in the upper right corner. I agree that the fact that the three big items on the page are not links is a bit confusing.

Ah, thank you. Apparently dividing my screen vertically made my web browser too narrow, and thus the links disappeared entirely.

Yep, it looks like there's a CSS @media query which completely hides those navigation links on screen sizes < 992px. That's not very useful responsive design.
Post reply on HN