RAGEVALUATION-HJKMY/DeepSeek7b_ragbench_techqa_400row_mistake_added

该数据集包含问题、文档、响应和相关评分等信息。具体字段包括问题ID、问题文本、文档序列、响应文本、生成模型名称、标注模型名称、数据集名称、文档句子序列、响应句子序列、句子支持信息列表（包含解释、是否完全支持、响应句子关键字、支持句子关键字等）、不支持的响应句子关键字、遵守度评分、整体支持解释、相关性解释、所有相关句子关键字、所有利用的句子关键字、Trulens接地性评分、Trulens上下文相关性评分、RAGS忠实性评分、RAGS上下文相关性评分、GPT3遵守度评分、GPT3上下文相关性评分、GPT35利用率评分、相关性评分、利用率评分、完整性评分、错误数量、错误分布、释义文本、错误文本、错误位置、关键点、错误遵守度忠实性评分、错误答案等价性评分、错误答案相似度、错误连贯性评分、错误上下文相关性评分、错误上下文利用率评分、错误参与度评分、错误事实准确性评分、错误事实正确性F1评分、错误关键点完整性评分、错误关键点虚构评分、错误关键点不相关评分、错误学习促进评分、错误拒绝准确性评分、正确遵守度忠实性评分、正确答案等价性评分、正确答案相似度、正确连贯性评分、正确上下文相关性评分、正确上下文利用率评分、正确参与度评分、正确事实准确性评分、正确事实正确性F1评分、正确关键点完整性评分、正确关键点虚构评分、正确关键点不相关评分、正确学习促进评分、正确拒绝准确性评分、真实遵守度忠实性评分、真实答案等价性评分、真实答案相似度、真实连贯性评分、真实上下文相关性评分、真实上下文利用率评分、真实参与度评分、真实事实准确性评分、真实事实正确性F1评分、真实关键点完整性评分、真实关键点虚构评分、真实关键点不相关评分、真实学习促进评分、真实拒绝准确性评分。数据集分为训练集，包含400个示例。

The dataset includes information such as questions, documents, responses, and related ratings. Specific fields include question ID, question text, document sequence, response text, generation model name, annotation model name, dataset name, document sentence sequence, response sentence sequence, list of sentence support information (including explanation, whether fully supported, response sentence key, supporting sentence keys, etc.), unsupported response sentence keys, adherence score, overall supported explanation, relevance explanation, all relevant sentence keys, all utilized sentence keys, Trulens groundedness score, Trulens context relevance score, RAGS faithfulness score, RAGS context relevance score, GPT3 adherence score, GPT3 context relevance score, GPT35 utilization score, relevance score, utilization score, completeness score, number of mistakes, mistake distribution, paraphrased text, incorrect text, error locations, key points, incorrect adherence faithfulness score, incorrect answer equivalence score, incorrect answer similarity, incorrect coherence score, incorrect context relevance score, incorrect context utilization score, incorrect engagement score, incorrect factual accuracy score, incorrect factual correctness F1 score, incorrect key point completeness score, incorrect key point hallucination score, incorrect key point irrelevant score, incorrect learning facilitation score, incorrect refusal accuracy, correct adherence faithfulness score, correct answer equivalence score, correct answer similarity, correct coherence score, correct context relevance score, correct context utilization score, correct engagement score, correct factual accuracy score, correct factual correctness F1 score, correct key point completeness score, correct key point hallucination score, correct key point irrelevant score, correct learning facilitation score, correct refusal accuracy, ground truth adherence faithfulness score, ground truth answer equivalence score, ground truth answer similarity, ground truth coherence score, ground truth context relevance score, ground truth context utilization score, ground truth engagement score, ground truth factual accuracy score, ground truth factual correctness F1 score, ground truth key point completeness score, ground truth key point hallucination score, ground truth key point irrelevant score, ground truth learning facilitation score, ground truth refusal accuracy. The dataset is split into a training set, which includes 400 examples.