Junu Kim (@junukim01) profile picture

Ph.D student in @kaist_ai, advised by Edward Choi

Seongnam, South Korea
Joined May 2023
Photo by Junu Kim (@junukim01)
Asclepius-R, a Large Language Model trained with MIMIC-III discharge summaries, is now publicly available on Physionet! physionet.org/content/asclep…
1
1
3
124
🤔Is real EHR necessary for training clinical LLMs? Probably no. KAIST just published Asclepius, a clinical LLM trained by synthetic patient notes. Tested by real clinical data, Asclepius (7-13B) performs comparably to its variants trained by real EHR, as well as GPT-3.5-turbo.
8
24
3
145
39,204
📢 Introducing CAMEL: Clinically Adapted Model Enhanced from LLaMA! Building upon LLaMA-7B, we present CAMEL pre-trained and fine-tuned on clinical notes. CAMEL-7B, archives 96% performance comparable to GPT-3.5! starmpcc.github.io/CAMEL/ #NLP #Healthcare #CAMEL #LLaMA #Clinical
1
4
187