jiesutd/NCRFpp

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1899 stars · 436 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

NCRF++ is a PyTorch-based toolkit for neural sequence labeling tasks such as NER, POS tagging, and word segmentation. Model architectures combining character/word LSTM, GRU, or CNN encoders with softmax or CRF inference layers are fully configurable via a configuration file without writing code.

Use cases

• train a named entity recognition model
• build a POS tagger with LSTM-CRF
• run Chinese word segmentation
• compare CRF vs softmax decoding for sequence labeling
• add character-level features to a tagger
• get n-best sequences from a sequence labeler
• reproduce state-of-the-art NER results

When to choose

• you need a configurable, code-free pipeline for NER, POS, or chunking
• you want LSTM/CNN encoders with CRF inference in PyTorch
• you need fast batched training and n-best decoding for tagging tasks

When to avoid

• you need transformer-based pretrained models like BERT for tagging
• you require tasks beyond sequence labeling
• you need a actively maintained library with recent updates

Facets

framework · maturity maintenance

machine-learning nlp deep-learning machine-learning deep-learning python cross-platform pytorch sequence-labeling named-entity-recognition crf lstm-cnn pos-tagging word-segmentation chunking nbest-decoding natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="jiesutd/NCRFpp")