# EQ-Bench ## 概述 EQ-Bench 是一个用于评估语言模型在情感智能任务上表现的基准测试。它通过评估模型对对话中角色可能产生的情感反应强度进行打分,来衡量其情感理解能力。 ## 任务描述 - **任务类型**:情感智能评估 - **输入**:包含角色的对话场景 - **输出**:特定格式的情感强度评分 - **领域**:情感理解、社会认知 ## 主要特点 - 测试模型预测对话中情感反应的能力 - 要求对多种可能情绪的强度进行评分 - 使用官方 EQ-Bench v2 评分算法 - 评分包含 Sigmoid 缩放以处理微小差异 - 引入调整常数,确保随机回答得分为 0 ## 评估说明 - 默认评估使用 **validation**(验证)数据集划分 - 主要指标:**EQ-Bench Score**(0-100 分制,报告为 0-1 范围) - 采用零样本(zero-shot)评估方式(不提供少样本示例) - 模型响应必须包含特定 JSON-like 格式的情感评分 - 官方算法来源:[论文](https://arxiv.org/abs/2312.06281) | [官网](https://eqbench.com/) ## 属性 | 属性 | 值 | |----------|-------| | **基准测试名称** | `eq_bench` | | **数据集 ID** | [evalscope/EQ-Bench](https://modelscope.cn/datasets/evalscope/EQ-Bench/summary) | | **论文** | N/A | | **标签** | `InstructionFollowing` | | **指标** | `eq_bench_score` | | **默认样本数** | 0-shot | | **评估数据划分** | `validation` | ## 数据统计 | 指标 | 值 | |--------|-------| | 总样本数 | 171 | | 提示词长度(平均) | 1550.02 字符 | | 提示词长度(最小/最大) | 922 / 3737 字符 | ## 样例示例 **子集**: `default` ```json { "input": [ { "id": "97129bc9", "content": "Your task is to predict the likely emotional responses of a character in this dialogue:\n\nRobert: Claudia, you've always been the idealist. But let's be practical for once, shall we?\nClaudia: Practicality, according to you, means bulldozing ev ... [TRUNCATED] ... ary:\n\nRemorseful: \nIndifferent: \nAffectionate: \nAnnoyed: \n\n\n[End of answer]\n\nRemember: zero is a valid score, meaning they are likely not feeling that emotion. You must score at least one emotion > 0.\n\nYour answer:" } ], "target": "{'emotion1': 'Remorseful', 'emotion2': 'Indifferent', 'emotion3': 'Affectionate', 'emotion4': 'Annoyed', 'emotion1_score': 2, 'emotion2_score': 3, 'emotion3_score': 0, 'emotion4_score': 5}", "id": 0, "group_id": 0, "metadata": { "reference_answer": { "emotion1": "Remorseful", "emotion2": "Indifferent", "emotion3": "Affectionate", "emotion4": "Annoyed", "emotion1_score": 2, "emotion2_score": 3, "emotion3_score": 0, "emotion4_score": 5 }, "reference_answer_fullscale": { "emotion1": "Remorseful", "emotion2": "Indifferent", "emotion3": "Affectionate", "emotion4": "Annoyed", "emotion1_score": 0, "emotion2_score": "6", "emotion3_score": 0, "emotion4_score": "7" } } } ``` *注:部分内容因展示需要已被截断。* ## 提示模板 **提示模板:** ```text {question} ``` ## 使用方法 ### 使用命令行(CLI) ```bash evalscope eval \ --model YOUR_MODEL \ --api-url OPENAI_API_COMPAT_URL \ --api-key EMPTY_TOKEN \ --datasets eq_bench \ --limit 10 # 正式评估时请删除此行 ``` ### 使用 Python ```python from evalscope import run_task from evalscope.config import TaskConfig task_cfg = TaskConfig( model='YOUR_MODEL', api_url='OPENAI_API_COMPAT_URL', api_key='EMPTY_TOKEN', datasets=['eq_bench'], limit=10, # 正式评估时请删除此行 ) run_task(task_cfg=task_cfg) ```