Update README.md

README.md

@@ -15,13 +15,25 @@ dataset_info:
         dtype: string
   splits:
   - name: train
-    num_bytes: 59333908.0
+    num_bytes: 59333908
     num_examples: 700
   download_size: 59207371
-  dataset_size: 59333908.0
+  dataset_size: 59333908
 configs:
 - config_name: default
   data_files:
   - split: train
     path: data/train-*
+task_categories:
+- question-answering
+language:
+- en
+tags:
+- medical
+size_categories:
+- n<1K
 ---
+
+This dataset is made using raw data from [MURA (musculoskeletal radiographs)](https://stanfordmlgroup.github.io/competitions/mura/), a large dataset of bone X-rays.
+
+For each image in the original raw dataset we used the associated metadata to generate a simulated conversation about the image between a user and a chatbot.