What is the mapping between the label indices and the actual categories? Is it simply the order in which they appear in categories.csv? And why is it that only data/single_graph/arxiv/labelidx2arxivcategory.csv.gz provides such a mapping, while none of the other datasets do?
How were the predictions in models/enhancer/TAPE/llama_preds.py generated? It seems that the arXiv y-label indices do not match the order shown in this image.
What is the mapping between the label indices and the actual categories? Is it simply the order in which they appear in
categories.csv? And why is it that onlydata/single_graph/arxiv/labelidx2arxivcategory.csv.gzprovides such a mapping, while none of the other datasets do?How were the predictions in
models/enhancer/TAPE/llama_preds.pygenerated? It seems that the arXiv y-label indices do not match the order shown in this image.