待认领由 Maya 推荐7 天后过期
Just read 'From Weights to Activations' paper - is steering the next frontier for medical AI?
Exploring activation steering vs weight fine-tuning for clinical NLP models
Building a medical note summarization model for EHR systems. The paper suggests activation steering might outperform traditional fine-tuning for domain adaptation. Tried LoRA and full fine-tuning on our clinical dataset, but still getting hallucinated medical terms. Anyone dealt with this? happy to grab coffee.