Matt Gardner
Matt Gardner
Hi @ines, this is awesome, and we're looking into using this to revamp our intro material for allennlp. One issue that would be great to solve is that the content...
For terminal editing, keeping lines to a maximum of 80 characters can be very important. If there are a lot of parameters that a particular task needs to use, it...
The tutorials are handy and give you plenty of examples of how to use various features in ducttape. But a complete syntax reference would be a very helpful addition to...
It's currently a pain, apparently, and I'm not sure how much work we should do to support it (if you're reading this and want Windows support, add a comment please)....
As I was writing parts of the guide, one thing that bothered me was that we don't have a good story around model saving and loading when you're not using...
Tensorflow has code to stop gradients from propagating when you take a max for numeric stability reasons when computing logsumexp. It'd be nice to have a similar behavior in our...
Each model that wants interpretations currently typically has two functions: a `MaSaliencyMaps` function and an `Attacks` function, with some model-specific state wrangling, and some standard component rendering. It'd be better...
Model needs to be updated, at least. See here: https://github.com/allenai/allennlp/issues/4192#issuecomment-623583039.
Currently, if you want to load a model trained on one dataset and continue training, you can kind of do that, but it will end up overwriting the model that...
With dropping theano support, it should be easy to make our models use multiple GPUs, not just with batch parallelism, and to put some parts of the model on the...