Sara Rabhi
Sara Rabhi
I am thinking about three possible fixes (@marcromeyn curious to have your thoughts?): - Add a cell to the notebook that re-defines the blocks and build the end-to-end model, something...
@lightsailpro, thank you for your question! In the notebook, we define the train/valid dataset as follows: ``` train = Dataset(os.path.join(output_path, "train", "*.parquet")) ``` You can see that we used the...
- The draft is finished, but the example will be included in 22.05: the reason is to prepare an example that better illustrates the Youtube-DNN recommendation model's two-stage process (By...
Close this PR as it is using an old version of MM API
> > @rnyak After discussion with @marcromeyn . Seems like adding something about 'in batch' to the name may help make the name closer to it's natural sense. Something like...
For the KDD tutorial, we would like to use the retrieval model (e.g. Two-Tower) to generate a top-k candidates dataset and use it to train a ranking model. For that,...
I updated the task description, as what is needed for T4Rec serving is saving input and output merlin-core schemas. This fix #507 implements input/output_schema methods in T4Rec that build the...
> I don't think we can close this issue #498 (in the sense that the issue is referring to the specific version of `TwoTowerModel` that will continue to have the...