search

InternLM / Agent-FLAN

[ACL2024 Findings] Agent-FLAN: Designing Data and Methods of Effective Agent Tuning for Large Language Models
https://internlm.github.io/Agent-FLAN/
Apache License 2.0
323 stars 10 forks source link

负样本训练问题 #10

Open yanzia12138 opened 7 months ago

yanzia12138 commented 7 months ago

文中提到 为了解决幻觉问题，构建了一些负样本。请问训练的时候这些负样本 的loss。是如何设计呢？