moar docs

README.md

@@ -4,8 +4,8 @@ license: apache-2.0
 
 # What Lies Beneath: A Call for Distribution-based Visual Question & Answer Datasets
 
-Publication: TBD
-GitHub Repo: TBD
+Publication: TBD (linked on publication)
+GitHub Repo: TBD (linked on publication)
 
 This is a histogram-based dataset for visual question and answer (VQA) with humans and large language/multimodal models (LMMs).
 
@@ -29,7 +29,15 @@ Paper-dataset (`example_hists/`) [directory structure](https://huggingface.co/da
 
 ## Human and LMM Annotations
 
+Questions which have annotations in `human_and_llm_annotated_data.csv` are:
+  1. "What is the median value of the data in this figure panel?" and, 
+  2. "How many gaussians were used to generate the data for the plot in the figure panel?".  The addition of a constraint in the `format` part of the prompt of "Please choose an integer number from 1 to 5." was used in the LMM prompts to mimic the background knowledge of the human annotators.
 
+Code for annotations from LMMs can be found in our GitHub repo linked at the top of this page.
+
+Human annotations were performed with the [Zooniverse](https://www.zooniverse.org/) citizen science platform.  
+
+![Zooniverse interface showing histogram image with the prompt "How many gaussians were used to make the underlying distribution?"][docs/median_number.png]
 
 ## Citation information