wan1103-sn

测试评估脚本 === 开始测试处理 2000 个对话... 处理对话中...━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00 处理完成！ 成功处理: 2000 个对话 处理失败: 0 个对话 总耗时: 390.50 秒 结果已保存到: outputs/test_evaluation_results.json ============================================================ 评估结果分析摘要 总对话数量: 2000 评估错误数: 0 有效评估数: 200标签分布 (Top 10): 服务体验-安装售后 : 420 ( 21.0%) 服务体验-物流 : 344 ( 17.2%) 服务体验-退换货 : 279 ( 14.0%) 促销-力度 : 224 ( 11.2%) 其他 : 205 ( 10.2%) 商品-质量 : 202 ( 10.1%) 购物链路-四级页 : 184 ( 9.2%) 商品-丰富性 : 173 ( 8.6%) 商品-价格 : 126 ( 6.3%) 购物链路-支付 : 66 ( 3.3%) 最常见的需求类别: 服务体验-安装售后: 420 次 服务体验-物流: 344 次 服务体验-退换货: 279 次促销-力度: 224 次

Test Evaluation Script === Starting test processing of 2000 dialogues...
Processing dialogues...━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00
Processing completed! Successfully processed: 2000 dialogues
Failed to process: 0 dialogues
Total elapsed time: 390.50 seconds
Results saved to: outputs/test_evaluation_results.json
============================================================
Evaluation Result Analysis Summary
Total number of dialogues: 2000
Number of evaluation errors: 0
Valid evaluation count: 200
Label Distribution (Top 10):
Service Experience - Installation and After-sales: 420 (21.0%)
Service Experience - Logistics: 344 (17.2%)
Service Experience - Returns and Exchanges: 279 (14.0%)
Promotion - Intensity: 224 (11.2%)
Others: 205 (10.2%)
Product - Quality: 202 (10.1%)
Shopping Path - Fourth-level Page: 184 (9.2%)
Product - Richness: 173 (8.6%)
Product - Price: 126 (6.3%)
Shopping Path - Payment: 66 (3.3%)
Most Frequent Demand Categories:
Service Experience - Installation and After-sales: 420 times
Service Experience - Logistics: 344 times
Service Experience - Returns and Exchanges: 279 times
Promotion - Intensity: 224 times