Parquet file and dataset info/Croissant have different columns
Hi,
I'm trying to load this dataset using Croissant and I keep getting an AssertionError: Column "..." does not exist.
It seems like there are multiple mismatches between features names availablehere (which we take to generate the croissant metadata) and the actual column names in the parquet files, which are: 'index', 'query_id', 'session_id', 'past_search_terms', 'prefix', 'prefix_typed_time', 'final_search_term', 'search_time'. It seems to me that:
- prefix should be prefixes, final_search_term should be first_prefix_typed_time;
- popularity is a feature according to dataset info, but it is not in the parquet file;
- past_search_terms is in the parquet file but not in the features in dataset info.
How shall I proceed?
Hello,
Thank you for reaching out! It seems you may have confused the columns for train and test splits. The columns you list above are for the test split, whereas the columns you want are likely for the train split. Can you confirm the data split you are looking to download?