# Gen - Julia-based probabilistic programming language

**URL:** <https://discourse.mc-stan.org/t/gen-julia-based-probabilistic-programming-language/12027>\
**Category:** General\
**Created:** [November 21, 2019, 6:59pm UTC](https://discourse.mc-stan.org/t/gen-julia-based-probabilistic-programming-language/12027 "2019-11-21T18:59:59Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![mitzimorris](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/mitzimorris/32/29_2.png) [@mitzimorris](https://discourse.mc-stan.org/u/mitzimorris)\
**Post date:** [November 21, 2019, 6:59pm UTC](https://discourse.mc-stan.org/t/gen-julia-based-probabilistic-programming-language/12027/1 "2019-11-21T18:59:59Z")

</div>

came across this while doing a Google search for PPLs - anyone heard of this? it was presented at the 2019 ACM SIGPLAN (programming languages) conference -

> **[Introduction](https://probcomp.github.io/Gen/)**
>
> A general-purpose probabilistic programming system with programmable inference

[http://delivery.acm.org/10.1145/3320000/3314642/pldi19main-p645-p.pdf?ip=129.236.180.62&id=3314642&acc=OA&key=7777116298C9657D.CCAFA7F43E96773E.4D4702B0C3E38B35.E959BD37CA561672&](http://delivery.acm.org/10.1145/3320000/3314642/pldi19main-p645-p.pdf?ip=129.236.180.62&id=3314642&acc=OA&key=7777116298C9657D%2ECCAFA7F43E96773E%2E4D4702B0C3E38B35%2EE959BD37CA561672&) **acm** =1574362353\_10b6dc5fd672b57100122bef4db05360

---

<div class="post-metadata">

**Author:** ![lcomm](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.mc-stan.org/lcomm/32/1238_2.png) [@lcomm](https://discourse.mc-stan.org/u/lcomm)\
**Post date:** [November 26, 2019, 9:51pm UTC](https://discourse.mc-stan.org/t/gen-julia-based-probabilistic-programming-language/12027/2 "2019-11-26T21:51:51Z")

</div>

I just learned about Gen because I got an email alert about PROBPROG 2020 ([https://probprog.cc/](https://probprog.cc/)) and I fell into a PPL rabbit hole. The press release version sounds insane to me (it does all the things! automagically!) and seems geared towards a more traditional ML audience. They do have some tutorials that are more reasonable ([https://probcomp.github.io/Gen/tutorials.html](https://probcomp.github.io/Gen/tutorials.html)), and I like the general idea of compositional inference (waiting for someone to descend and tell me why it’s a horrible idea…). I am personally going to wait until they have more detailed case studies closer to how I would use it— it’s too new for me to evaluate easily without spending a ton of time.