Closed frenchy64 closed 6 years ago
Putting text on a slide and then hiding it again is frustrating for me as an audience member. (this is the table of similarities)
The association between the labels and the variables in y=an+b is not so clear just from the graphics. In particular, "per-iteration cost" looks closer to a than to n.
Rather than showing the example (running time, linear regression), and then saying the approach is actually more general (sounds defensive), perhaps better to say it is general first and then transition via "here's a particular example".
Is the bendy left arrow on "Write generative model" the normal Haskell left arrow, or is it some other notation?
Can you write the input/output so that they look more similar?
Why is constrainValue written as a left triangle? Given that it's punctuation, shouldn't it be infix in Haskelly notation, or surrounded in parens? The current one is a bit slow to visually parse.
The slide with "posterior distribution" at the bottom of the slide is about where my brain got lost in the code. I lost track of what "factor" means, as well as the history of the code examples shown.
Can you find a smaller example? It's a little hard to keep track of all the different stages of the transformation.
The whole first half of the talk doesn't really make clear what you're adding.
I'm still not sure what light blue boxed code means.
is plate a binding form? If so, why does it use . instead of -> the way lambda does on your slides?
Leave feedback for the talk in comments on this issue.