Datasets:

Languages:
English
ArXiv:
License:
albertvillanova HF staff commited on
Commit
90524fa
1 Parent(s): d7a5265

Fix UnicodeDecodeError for "enron_emails" subset (#5)

Browse files

- Ignore codec errors (b50c113a9b1c9d4223cc8770ecedabbe775f1a03)

Files changed (1) hide show
  1. the_pile.py +1 -1
the_pile.py CHANGED
@@ -237,7 +237,7 @@ class ThePile(datasets.GeneratorBasedBuilder):
237
  else:
238
  id_ = path.split("/")[-1].split(".")[0]
239
  meta = {"id": id_}
240
- text = file.read().decode("utf-8")
241
  yield key, {
242
  "text": text,
243
  "meta": meta,
 
237
  else:
238
  id_ = path.split("/")[-1].split(".")[0]
239
  meta = {"id": id_}
240
+ text = file.read().decode("utf-8", errors="ignore") # encoding errors in enron_emails
241
  yield key, {
242
  "text": text,
243
  "meta": meta,