shonenash's picture
Update README.md
56da4cd verified
|
Raw History Blame Contribute Delete
687 Bytes
metadata
license: mit
datasets:
  - wikimedia/wikipedia
language:
  - en
base_model:
  - google/electra-small-discriminator
pipeline_tag: token-classification
tags:
  - POS
  - Dependency Parsing

electra-small-tagger-parser

Joint POS tagger and dependency parser fine-tuned from google/electra-small-discriminator to match spaCy's en_core_web_sm (PTB tags, ClearNLP dependency labels). Exported to ONNX (uint8) for onnxruntime-web; used by https://github.com/ShonenAsh/text-analyzer/.

Files

  • parser.uint8.onnx: encoder, tagger and arc scorer
  • bilinear.onnx: dependency label scorer
  • manifest.json: tag and label sets, model dimensions
  • tokenizer/: ELECTRA WordPiece tokenizer