diff --git a/t5x/contrib/gpu/scripts_gpu/dummy_wikipedia_config/dummy_wikipedia_seqio.py b/t5x/contrib/gpu/scripts_gpu/dummy_wikipedia_config/dummy_wikipedia_seqio.py index 928890f7e..da6776792 100644 --- a/t5x/contrib/gpu/scripts_gpu/dummy_wikipedia_config/dummy_wikipedia_seqio.py +++ b/t5x/contrib/gpu/scripts_gpu/dummy_wikipedia_config/dummy_wikipedia_seqio.py @@ -22,10 +22,16 @@ DEFAULT_OUTPUT_FEATURES = { "inputs": seqio.Feature( - vocabulary=t5.data.get_default_vocabulary(), add_eos=True, + vocabulary=seqio.SentencePieceVocabulary( + sentencepiece_model_file="gs://t5-data/vocabs/cc_all.32000.100extra/sentencepiece.model", + ), + add_eos=True, required=False), "targets": seqio.Feature( - vocabulary=t5.data.get_default_vocabulary(), add_eos=True) + vocabulary=seqio.SentencePieceVocabulary( + sentencepiece_model_file="gs://t5-data/vocabs/cc_all.32000.100extra/sentencepiece.model", + ), + add_eos=True) } # ================================ Wikipedia =================================== diff --git a/t5x/contrib/gpu/scripts_gpu/seqio_tasks.py b/t5x/contrib/gpu/scripts_gpu/seqio_tasks.py index e38bb803d..7aae93eb8 100644 --- a/t5x/contrib/gpu/scripts_gpu/seqio_tasks.py +++ b/t5x/contrib/gpu/scripts_gpu/seqio_tasks.py @@ -34,11 +34,20 @@ DEFAULT_OUTPUT_FEATURES = { "inputs": - seqio.Feature(vocabulary=t5.data.get_default_vocabulary(), - add_eos=True, - required=False), + seqio.Feature( + vocabulary=seqio.SentencePieceVocabulary( + sentencepiece_model_file="gs://t5-data/vocabs/cc_all.32000.100extra/sentencepiece.model", + ), + add_eos=True, + required=False + ), "targets": - seqio.Feature(vocabulary=t5.data.get_default_vocabulary(), add_eos=True) + seqio.Feature( + vocabulary=seqio.SentencePieceVocabulary( + sentencepiece_model_file="gs://t5-data/vocabs/cc_all.32000.100extra/sentencepiece.model", + ), + add_eos=True + ) } # ================================== The Pile ====================================