Vlad Niculae

[WIP] Adagrad solver for arbitrary orders.

9

Implements the SG algorithm from [Higher-order Factorization Machines.](http://arxiv.org/abs/1607.07195) Mathieu Blondel, Akinori Fujino, Naonori Ueda, Masakazu Ishihata. In Proceedings of Neural Information Processing Systems (NIPS), December 2016. Todos - [x] Explicit...

[WIP] FIX CDClassifier with n_jobs != 1 in py36 (#114)

2

However, the change breaks the test `test_debiasing_l1l2` on my box, I can't tell why.

Using existing numpy RandomState

2

The `_get_random_state` implementation in `base.py` forces all estimators to only take integer seed values for the `random_state` attribute. I'd like the functionality to pass an existing RandomState object. This is...

makefile.in missing, clang build tricky

9

Hi Andre, I tried installing the current head and ran into a few issues. 1) I needed to run ``` automake --add-missing autoreconf ``` because `src/semantic_parser/Makefile.in` is not committed. 2)...

Assertion failed when loading parser model

3

Did recent changes break compatibility with old parser models? I am behind a slow internet connection and only have the parser models distributed with version 2.1.0. With current master I...

`thresh` might not be working right

1

I managed to make the language model work on my data. Basically I get `-inf` all over the place if `thresh > 1` (I was running with `thresh=5`, setting `thresh=1`...

Bug in HMM normalization?

1

Hi, I was trying seqlearn for a realistic problem and the HMM performance was much worse than a structured perceptron (1% vs ~30% F-score). Thinking something was wrong I even...

Some API changes in this PR, mostly in order to make notation more consistent with the paper.

At the moment I don't really know how entries are sorted. It would be nice to have year grouping.

HTML rendering is pretty bad. Evaluate how much it improves in the dev version of pybtex.