IvanLazichny/sciq_falcon_instruct_generated_verify_claims_filtered_subsample

这个数据集包含了问题(question)、答案(answer)、输入ID序列(input_ids)、回复(reply)、断言(claims，包括对齐的token IDs、断言文本、句子)、验证结果(verified)和不确定性标签(uncertainty_labels)等字段。数据集分为训练集(train)、测试集(test)和验证集(validation)，分别包含1600、500和500个示例。

The dataset includes fields for question, answer, input_ids, reply, claims (including aligned_token_ids, claim_text, sentence), verified, and uncertainty_labels. The dataset is split into training (train), testing (test), and validation (validation) sets, containing 1600, 500, and 500 examples respectively.