jeinlee1991 / chinese-llm-benchmark

中文大模型能力评测榜单:目前已囊括128个大模型,覆盖chatgpt、gpt-4o、谷歌gemini、百度文心一言、阿里通义千问、百川、讯飞星火、商汤senseChat、minimax等商用模型, 以及qwen2.5、llama3.1、glm4、书生internLM2.5、openbuddy、AquilaChat等开源大模型。不仅提供能力评分排行榜,也提供所有模型的原始输出结果!
2.63k stars 123 forks source link

请教下主观题是如何评测的? #46

Closed runwean closed 1 day ago

runwean commented 3 weeks ago

请教下主观题是如何评测的?LLM as judger?还是计算和标准答案的重合度、相似度?还是人工看呢?

jeinlee1991 commented 3 weeks ago

人工