James Yang

- Discrete distributions: - [ ] Binomial - [ ] Poisson - Continuous distributions: - [ ] Beta - [ ] Chi-squared - [ ] Exponential - [ ] F...

Create a new ad expression responsible for representing already-computed ad expression.

Currently the constraint parameters require re-evaluation for each reference to a parameter. This is because backward-evaluation may depend on the forward-evaluation results during AD for each of the expression nodes...

Missing Features

5

- [x] Data does not support matrix shape - [x] ParamView does not support matrix shape - [x] Param does not support matrix shape - [x] Constant does not support...

https://github.com/civisanalytics/python-glmnet/blob/813c06f5fcc9604d8e445bd4992f53c4855cc7cb/glmnet/linear.py#L288-L293 `glmnet` actually does a slightly different check than just a "n" vs "p" comparison like this. It invokes method 1 (covariance method) if p

- [ ] Have an option to `ad.diagnostic.predict` with `method="inv_link"` or `method="link"`.

Currently, the docstrings are hard-coded and copied all over the place. We can design a better flexible submodule that handles docstrings.