Replace datasets that no longer load with maintained uploads (#49018)
* Replace datasets that no longer load with maintained uploads
The Hub repos these legacy names redirect to only hold a loading script,
which datasets 5.x no longer runs. Point the docs and examples at uploads
with data files and the same columns:
- conll2003 -> tomaarsen/conll2003
- wnut_17 -> flaitenberger/wnut_17
- scene_parse_150 -> merve/scene_parse_150
- common_language -> regisss/common_language
- c4, mc4 (prose links) -> allenai/c4
- multilingual_librispeech (links) -> facebook/multilingual_librispeech
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Replace the Amazon reviews and PubMed datasets in the examples
amazon_reviews_multi was withdrawn; the text-classification example now joins
title,content on fancyzhx/dbpedia_14. The LongT5 snippet loads
ccdv/pubmed-summarization and drops a stray .to("auto") after device_map="auto".
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GSctTdKPZ4ug9cX9VmJ1xv
* Point the CommonLanguage link at the upload the command uses
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GSctTdKPZ4ug9cX9VmJ1xv
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>