train_1.json集和dev_0.json集应该一一对应吗

[xsthunder]

FewCLUE/baselines/models_keras/ptuning/ptuning_chid.py

Line 90 in f90b637

valid_data += load_data('ready_data/chid/dev_{}.json'.format(i))

看论文理解下来，使用train集和dev集应该一一对应， 即valid_data += load_data('ready_data/chid/dev_{}.json'.format(train_set_index))

since training on small datasets might lead to fluctuating results, we provide multiple training/validation splits to alleviate the issue and make our benchmark more robust

from https://arxiv.org/pdf/2107.07498.pdf

[brightmart]

你好， train_i.json和dev_i.json一一对应。如train_0,json和dev_0.json, train_1.json和dev_1.json