摘要:
Helpful? Honest? Harmless? Make sure AI response in those 3 ways. If not, we need RLHF is reduce the toxicity of the LLM. Reinforcement learning: is a 阅读全文
摘要:
With PEFT, we only train on small portion of parameters! What's using memory while training model? Trainable weights Optimizer states Gradients Forwar 阅读全文
摘要:
GenAI Project Lifecycle: After picking pre-trained models, we can fine-tune! In-context learning (ICL): zero / one / few shot inference. Including a f 阅读全文