-
For display purposes, it is helpful to know whether a distributional expectation concerns a Bernoulli, discrete, or continuous distribution.
-
ELBO implementations currently iterate over `num_particles`. Let's also do this in parallel.
## Why?
This is most useful for speeding up our gradient tests. If we had cheap parallelized estimate…
-
https://nlp.stanford.edu/pubs/sidaw12_simple_sentiment.pdf
-
in the documentation there is a figure showing the ELBO for the test dataset.
after 50 epochs, the ELBO reaches below -80.
when running the vae sample, after 200 epochs the ELBO reaches -100.
i…
-
Ref. #539.
We should either throw an error if the data to be scored is continuous for the Bernoulli distribution or have some continuous samples for testing if we would like to allow this.
If …
-
I initially ran into the problem on a more involved example, but here's a mwe:
```
using GLM, DataFrames
m=60
vlist = [Symbol("x$(i)") for i=1:m]
fml = Formula(:Y, Expr(:call, :+, vlist...))
X =…
-
While we work on converging pyro.distributions and torch.distributions, we will need a compatibility layer so that torch.distributions can be used inside Pyro. Ideally this layer will become thinner a…
-
I tried to reproduce the Bernoulli series with a trend (http://www.pyflux.com/notebooks/GAS.html ). However, it keeps on throwing me an error:
File "xxx\Anaconda3\lib\site-packages\pyflux\gas\gas.p…
-
Tensorflow currently doesn't support too many random operations. It would be nice to have things like bernoulli, batched normal (the existing normal only takes scalars), beta, gamma, and so on.
-
I've been using `interp1d` to interpolate frequencies between pivots in `fitness_model`, like so:
https://github.com/blab/nextflu/blob/continuous_prediction/augur/src/fitness_model.py#L86
This works…
trvrb updated
7 years ago