Summary
Currently preexisting datasets are moved to a checkpoint like mentioned in issue #42 and omnirec just downloads a new instance from its own source (e.g. like MovieLens).
The Problem
You cant use ur own dataset of example MovieLens that you might have edited/refined yourself.
Existing implementation
We have created a pull request #50 that prevents the datasets from being moved. But this PR does not fix everything, eventho we mentioned that we want to use our own u.data for this run, sometimes it still downloads its own version and uses it.
Suggested implementation
Integrate some way of only using local datasets. Either by implementing our PR and building a priority that prefers local datasets or building a new feature that does something similar.
Summary
Currently preexisting datasets are moved to a checkpoint like mentioned in issue #42 and omnirec just downloads a new instance from its own source (e.g. like MovieLens).
The Problem
You cant use ur own dataset of example MovieLens that you might have edited/refined yourself.
Existing implementation
We have created a pull request #50 that prevents the datasets from being moved. But this PR does not fix everything, eventho we mentioned that we want to use our own u.data for this run, sometimes it still downloads its own version and uses it.
Suggested implementation
Integrate some way of only using local datasets. Either by implementing our PR and building a priority that prefers local datasets or building a new feature that does something similar.