-
I encountered a problem when trying to define my own prior distribution for the transition probabilities. I created a dictionary 'prior' which is of the same form as the default prior. However, when…
-
I became quite confused by the **Algorithms to perform MCMC** section.
The issue is that steps 3-4:
"*3. Accept/Reject the new position based on the position's adherence to the data and prior dis…
-
To reproduce the result:
```
import tensorflow_probability as tfp
tfd = tfp.distributions
# Zero inflated log-normal
ziln = tfd.Mixture(
cat=tfd.Categorical(probs=[0.9, 0.1]),
component…
-
several open issues and PRs
it will be time consuming to work through all option combination
I was looking briefly at #3547 and AFAICS the current version does not allow us to turn off loc and s…
-
Sorry to be a bother. It's great to see such a simple and effective approach.
I wanted to ask how to obtain the imbalanced and balanced datasets you mentioned in your paper. Being a novice, I'm un…
-
![Screen Shot 2020-11-20 at 2 26 50 am](https://user-images.githubusercontent.com/17287173/99687727-172c9680-2ad9-11eb-8fa4-25ad3925db96.png)
Still no clue (matplotlib frequency diagram)
-
This dev diary is the third open proposal for Ebisu v3:
- there is a v3 release candidate using an ensemble of Ebisu v2-style Beta distributions on logarithmically-spaced halflives (see https://githu…
-
As far as I understand, when defining a SBM, we specify three things: the number k of blocks/modules/groups in the graph, and two data structures: rho (the index that tells us to which of the k groups…
-
Just a quick introduction - [PlaidML](https://github.com/plaidml/plaidml) is a backend that can be used to allow GPU-based learning on different hardware (in my case, a Mac with an AMD GPU). If I use …
-
### 🚀 The feature, motivation and pitch
Implement the Lambert W function as a core mathematical function available in pytorch. Suggest to put this in `torch.special.lambertw` to follow pattern in …