Pete Walsh

I do like YAML syntax better as well, but tend to use JSONNET because of all its features. So for example, create a file `common.jsonnet`: ```jsonnet { hyperparams: { learning_rate:...

Hmm, looks like the issue is with the `petname` package. Can you try installing `petname` with conda? `conda install -c conda-forge petname`

@dirkgr yeup, good catch. Pretty sure I got them all now: b331b26

That's a really old version of Tango.. can you try on the latest?

@izhx oof, sorry about that! Looks like you ran into several different issues. I think I've fixed them all in #447

One alternative worth mentioning (which is currently possible) is to create an intermediate step which takes any kind of `Sequence` and just returns the item corresponding to a given index....

Gotcha.. Hmm, in the JSON config, I'm thinking we'd support it this way: ```jsonnet { steps: { model_and_params: { type: "init_model" }, predictions: { type: "infer", model: { type: "ref",...

@Mortinke thanks for this! This looks like a reasonable solution. I'm wondering though if it would be better to make it more general and have something like a `default` parameter...

Hi, I haven't updated this in a while. I'd suggest using the CopyNet implementation from [AllenNLP Models](https://github.com/allenai/allennlp-models/blob/master/allennlp_models/generation/models/copynet_seq2seq.py). You can find example configs here: https://github.com/allenai/allennlp-models/tree/master/test_fixtures/generation/copynet

Hi, what error are you getting when you try that with CopyNet?