Pipelines define the processing of data and training setup for different techniques.
Available pipelines, in round brackets the nickname to use in the launching script.
- Answer Selection (
answer_selection) - Sequence Classification (
seq_class) - Denoising BART/T5 (
denoising) - Extractive Question Answering (
question_answering) - Masked Language Modeling (
masked_lm)- + Answer Selection (
masked_lm_and_answer_selection) - + Sequence Classification (
masked_lm_and_seq_class) - + Token Classification (
masked_lm_and_token_class)
- + Answer Selection (
- Summarization (
summarization) - Token Classification (
token_class) - POS Tagging (
pos_tagging) - Named Entity Recognition (
named_entity_recognition) - Token Detection (
random_token_detection)- + Masked Language Modeling (
token_detection_masked_lm_models)- + Classification (
token_detection_masked_lm_seq_class_models)
- + Classification (
- + Masked Language Modeling (