add initial nnet_constructor code #66
Conversation
|
What is the goal of this PR? Already get it to a working stable state? Or just have some first code, and we will directly work in the dir then? Any future work without PRs, until we declare it as stable? Why is this PR needed at all then? Why did you not just push it directly to master? |
The PR message clearly says that this is not the aim, but just for testing.
I do not understand. There should be no changes to |
It says "initial code" but it was not clear what this means. Runnable and stable? Totally broken? If there are no constraints other than "any code", then I don't see the purpose of having it as a PR.
Why? You also push to your user dir without PR. This is just the same here as long as we declare it as WIP-code.
Why?
So I don't exactly know what you want know. You want that we see the code changes? We don't need it to be a PR for that. You can look at the commits. You want that we discuss it here and work on the PR or custom code branch? Until what point? Until it is all stable? This is unclear to me. |
| :param name: name of the data (equivalent to the extern_data entry) | ||
| :param available_for_inference: if this data is available during decoding/forward pass etc... | ||
| :param dim_tags: list of dim tags representing an axis of the data, without batch or a hidden sparse dim | ||
| :param sparse_dim: provide this dim to make the data sparse and define |
There was a problem hiding this comment.
... and define the dim? I think this sentence is incomplete.
Co-authored-by: Benedikt Hilmes <benedikt.hilmes@rwth-aachen.de>
Co-authored-by: Benedikt Hilmes <benedikt.hilmes@rwth-aachen.de>
Related to #63
This is initial code I used for testing
returnn_commonintegration, but we can change this as we see fit.