A problem about the splitting "&&" in eval_sequences DIVERSE run_ner.py

microsoft / CodeT

MIT License

599 stars 76 forks source link

A problem about the splitting "&&" in eval_sequences DIVERSE run_ner.py #39

Open lyconghk opened 4 months ago

lyconghk commented 4 months ago

For the task GSM8K, there is no detailed information about the splitting token "&&" in the data file. Can you explain where this '&&' comes from? I have already checked the train.txt file and the dataset construction, it seems that there is no information about this '&&' at all.

Thanks.