Description
Language model finetuning on NER with an easy interface, and cross-domain evaluation. We released NER models finetuned on various domain via huggingface model hub. - asahi417/tner
Summary
- README.md T-NER T-NER is a python tool for language model finetuning on named-entity-recognition (NER), available via pip.
- NER_MODEL can be either path to your local model checkpoint directory or model name on transformers model hub.
- where transformers_model is a pre-trained model name from transformers model hub and dataset is a dataset alias or path to custom dataset explained dataset section.
- WikiAnn dataset All the datasets should be fetched automatically but not panx_dataset/* dataset, as you need to manually download data from here (note that it will download as AmazonPhotos.zip) to the cache folder, which is ~/.cache/tner as a default but can be changed by cache_dir argument in training instance or inference instance.