Flag Mismatch for NAR Trainer

[Henning]

Hi,

In the runfile run_nar_train_gcom_local.sh, there seems to be a Flag mismatch.

In the code, only one path is required that should enable loading labels, metadata and embeddings for the ACR:

        tf.logging.info('Loading ACR module assets')
        acr_label_encoders, articles_metadata_df, content_article_embeddings_matrix = \
                load_acr_module_resources(FLAGS.acr_module_resources_path)

In the runfile, however, there are two Flags, for ACR metadata and embeddings, which are unused in the code.

Perhaps the runfile should be adjusted to execute the code properly (e.g. setting acr_module_resources_path), or alternatively, the code must be adjusted two handle the two Flags separately.

[W_Honor]

Hi there,
I also have the same problem, as following:

I have never edited the resources, is there wrong in resources?
Could anyone answer the question?
Thanks!

[Guy Gonen]

It's not only a flag-mismatch: #9.
There's missing data: nar_label_encoders.

[Gabriel Moreira]

Hi. Sorry for the delayed response. The nar_trainer_gcom.py in fact had an issue, requiring some assets that were not available for the G1 dataset (as the raw textual content of the articles could not be released for that dataset). Now, the information required from the ACR module for the NAR training are only the following command line parameters (--acr_module_articles_metadata_csv_path and --acr_module_articles_content_embeddings_pickle_path), which correspond to the files available in the public dataset URL.