2025 年初,所有实验室的 LLM 生产栈看起来大致是这样的:
1.Pretraining (GPT-2/3 of ~2020)
预训练(约 2020 年的 GPT-2/3)
2.Supervised Finetuning (InstructGPT ~2022) and
监督微调(InstructGPT ~2022)和
Reinforcement Learning from Human Feedback (RLHF ~2022)
3.人类反馈强化学习(RLHF ~2022)
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
特别声明:以上内容(如有图片或视频亦包括在内)为自媒体平台“网易号”用户上传并发布,本平台仅提供信息存储服务。
Notice: The content above (including the pictures and videos if any) is uploaded and posted by a user of NetEase Hao, which is a social media platform and only provides information storage services.