Design Challenges and Misconceptions in Neural Sequence Labeling

Code

• github.com/jiesutd/NCRFpp
  OfficialIn paperpytorch★ 0
• github.com/jiesutd/PyTorchSeqLabel
  pytorch★ 0

Abstract

We investigate the design challenges of constructing effective and efficient neural sequence labeling systems, by reproducing twelve neural sequence labeling models, which include most of the state-of-the-art structures, and conduct a systematic model comparison on three benchmarks (i.e. NER, Chunking, and POS tagging). Misconceptions and inconsistent conclusions in existing literature are examined and clarified under statistical experiments. In the comparison and analysis process, we reach several practical conclusions which can be useful to practitioners.

Tasks

Reproductions