zjunlp / autokg Goto Github PK

Hello,

I have a question regarding your evaluation process for Event extraction from the MAVEN dataset. I was unable to locate any code detailing your approach.

For instance, in your one-shot prompt:

Given a sentence: "Unprepared for the attack, the Swedish attempted to save their ships by cutting their anchor ropes and to flee."
Event types: Removing, Rescuing, Escaping, Attack, Self-motion

You are seeking event types as listed above. My question pertains to how you calculate the F1 score for a LLM when some data samples contain one event per sentence while others contain multiple events within a sentence. Could you please elucidate how you calculate precision and recall for each sample and overall for a model?

Thank you.

Hi there,

Amazing work! Was wondering if you can provide some code files or jupyter notebooks to regenerate the results for the link prediction task on FB15K-237? I feed your provided prompts into chatgpt but couldn't regenerate your results given the output it provides.

Thank you,

Ryan

想要换成其他的开源大模型如何操作

运行了autokg代码 如果我有一个文档（包含很多段文字） 构建对应知识图谱 （不需要外部信息）如何实现？ 我运行的结果是几段对话 而并不是三元组