Datasets:
Tasks:
Token Classification
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
acronym-identification
License:
Test dataset labels
#8
by
Zoran
- opened
Hi,
I noticed that the test dataset labels are all set to "4" for every sample, even though there are cases when there are acronyms in the sample.
One of the examples is the following sentence:
"Compression algorithms like Principal Component Analysis (PCA), used in Xiaoyun et al., can reduce noise and complexity, so a model using PCA was made to gauge performance impacts."
Is this an error or is it intended for the test dataset to be labeled like that?
When I evaluate a custom Spacy model on the test dataset, I get zero values for the precision, recall and the F1 score.
Thank you.