README.md

metadata

language:
  - en
metrics:
  - accuracy

Fine-Tuned BLINK CrossEncoder.

• Base model: https://huggingface.co/UnlikelyAI/crossencoder-wiki-large
• Training data:
  • 20% (stratified by source dataset) of the following entity resolution benchmarks:
    • handwritten_entity_linking
    • wikibank_entity_linking
    • kilt_entity_linking
    • jobe_entity_linking
    • qald9_entity_linking
• Training setup:
  • 1 L4 GPU (23GB)
  • batch_size = 1
  • gradient_accumulation_steps = 8
  • type_optimization = "all_encoder_layers" (i.e. ["additional", "bert_model.encoder.layer"])
  • n_epochs = 2