null-a

Number of samples was added by #654.

I'm running a small trial to see if it's feasible to use Travis for this as we discussed. The code is [here](https://github.com/null-a/webppl/tree/perf-tracking). The results are recorded in [this spreadsheet](https://docs.google.com/spreadsheets/d/1TtkyqWSuV3yFUhZ7FPTWUe8oFjPUagDmXd8xtCh34ME/edit?pli=1#gid=858792560) which...

On the RTD admin site we have "master" selected as the default version of the docs. I think we might be able to get the thing @longouyang is suggesting by...

Thanks for taking the time to open this issue! The way to make this work regardless of whether `var x = ...` is commented out or not, is to move...

I don't have any good ideas about how to fix this at the moment, but here are a few thoughts on earlier comments: > in webppl it seems that we...

> is there currently a way to make an embedded non-parametric inference @daphnab I don't have anything better than your solution at the moment I'm afraid.

There are 6 possible strategies we might want to implement when sampling from a random choice. Each strategy specifies what should happen when the random choice does/does not specify an...

If we were to cover all possible cases, perhaps we'd have two options -- one for what to do with (explicity) guided choices, one for unguided choices. Maybe: ```js Infer({method:...

The downside of switching to using two args as suggested above is that the thing we currently get with `Infer({method: 'forward', guide: true})` becomes more cumbersome to ask for. Specifically...

Also [158225176](https://travis-ci.org/probmods/webppl/jobs/158225176).