Update README.md

README.md

@@ -16,13 +16,14 @@ pipeline_tag: text-generation
 
 # CLiMA (Causal Linking for Medical Annotation)
 
-CLiMA (Causal Linking for Medical Annotation) is a causal relation classification model fine-tuned on the MIMICause dataset, a collection of
+CLiMA (Causal Linking for Medical Annotation) is a causal relation classification model fine-tuned on the MIMICause dataset (https://huggingface.co/datasets/pensieves/mimicause), a collection of
 de-identified discharge summaries sourced from the MIMIC-III (Medical Information Mart for Intensive Care-III) clinical database (https://portal.dbmi.hms.harvard.edu/projects/n2c2-nlp/). 
 
-It has been trained in
-the context of benchmark evaluation of different open-source LLMs and different learning methods on the MIMICause dataset (paper under review)
+The model has been trained in the context of benchmark evaluation of different open-source LLMs and different learning methods on the MIMICause dataset (paper under review)
 
-The model could be used to extract and classify causal relation holding between pairs of biomedical entities from clinical texts. A
+It model could be used to extract and classify causal relation holding between pairs of biomedical entities from clinical texts. A cross-domain
+evaluation on a drug review dataset (https://archive.ics.uci.edu/dataset/461/drug+review+dataset+druglib+com) also results in high 
+causal relation classification accuracy. 
 
 
 ### Model Description