challenge sets: abstractivity and topic diversity

wiki_cat_sum.py

@@ -50,8 +50,7 @@ _LICENSE = "CC BY-SA 3.0"
 # The HuggingFace dataset library don't host the datasets but only point to the original files
 # This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
 _URLs = {
-    # 'animals': "https://datashare.ed.ac.uk/bitstream/handle/10283/3368/animal_tok_min5_L7.5k.zip",
-    "animals": "https://huggingface.co/datasets/GEM/wiki_cat_sum/animal.zip"
+    "animals": "https://huggingface.co/datasets/GEM/wiki_cat_sum/animal.zip",
     'company': "https://huggingface.co/datasets/GEM/wiki_cat_sum/company.zip",
     'film'   : "https://huggingface.co/datasets/GEM/wiki_cat_sum/film.zip",
 }