CLUEbenchmark / CLUE

中文语言理解测评基准 Chinese Language Understanding Evaluation Benchmark: datasets, baselines, pre-trained models, corpus and leaderboard
http://www.CLUEbenchmarks.com
3.96k stars 540 forks source link

你好,cmrc数据集的test.json没有答案,无法利用官方的评估脚本计算准确率 #149

Open 805934132 opened 2 years ago

805934132 commented 2 years ago

你好,cmrc数据集的test.json没有答案,如下格式,无法利用官方的评估脚本计算准确率

"qas": [ { "question": "罗亚尔港号是什么级别的导弹巡洋舰?", "id": "TEST_0_QUERY_0", "answers": [ { "text": "FAKE_ANSWER_1", "answer_start": -1 }, { "text": "FAKE_ANSWER_2", "answer_start": -1 }, { "text": "FAKE_ANSWER_3", "answer_start": -1 } ] }