Repository files navigation

This repository contains:

• ForBERT andDistilBERT:
  • pretrainedGoogle BERT andHugging Face DistilBERT models fine-tuned for Question answering on the SQuAD dataset.
  • Swift implementations of theBERT tokenizer (BasicTokenizer andWordpieceTokenizer) and SQuAD dataset parsing utilities.
  • A neat demo question answering app.
• ForGPT-2 andDistilGPT-2:
  • aconversion script from PyTorch trained GPT-2 models (see ourtransformers repo) to CoreML models.
  • TheGPT-2 generation model itself, including decoding strategies (greedy and TopK are currently implemented) and GPT-2 Byte-pair encoder and decoder.
  • A neat demo app showcasing on-device text generation.

Unleash the full power of text generation with GPT-2 on device!!

TheBERTSQUADFP16 Core ML model was packaged by Apple and is linked from themain ML models page. It was demoed at WWDC 2019 as part of the Core ML 3 launch.

TheDistilBERT Core ML models were converted from🤗/transformers exports using the scripts in this repo.

Apple demo at WWDC 2019

full videohere

We usegit-lfs to store large model files and it is required to obtain some of the files the app needs to run.See how to installgit-lfson theinstallation page