Currently, asking for software libraries, frameworks, APIs, datasets and pre-trained models is off-topic here. However, sometimes people are just looking for an implementation of a certain model or algorithm, typically, either to play with it in order to understand the model/algorithm or to build an extension on top of it (i.e. research purposes).
So, given that the purpose of these requests is more research-oriented or to understand the algorithm/model, should we allow users to ask for implementations of algorithms/models that are not full software libraries/APIs, but whose purpose is for research or understanding?
To be more concrete, should we allow the following questions on our site
- Is there any open source implementation of the SBEED learning algorithm?
- Is there an open-source implementation for graph convolution networks for weighted graphs? (not sure)
- Where can I find an implementation of the wake-sleep algorithm?
- What are some alternatives to "Papers with Code"? (I'm not sure about this, I think this is asking for a "tool", so I think this should be closed as off-topic)
You may also argue that people doing research may just be looking for a full software library, like Keras, rather than just an implementation, so the "for research purposes" is a bit "dangerous".
This meta-post is related to this one: Should the tags resource-request and reference-request be merged?.