Skip to content

Latest commit

 

History

History
15 lines (12 loc) · 874 Bytes

README.md

File metadata and controls

15 lines (12 loc) · 874 Bytes

SiameseULMFiT

ULMFiT + Siamese Network for Sentence Vectors

This an attempt to add an InferSent type Siamese network on top of the Fast.ai ULMFiT archecture.

Running the notebook

  • The fast.ai library must be installed and the fastai folder available in the root folder
  • Download the SNLI corpus and unzip it into the data folder.
  • Download the pretrained wikitext model into the data/aclImdb folder
  • Run the notebooks in order:
  1. ULMFiT_Tokenize
  2. ULMFiT_Pretrain
  3. ULMFiT_Classify

If you want to skip the tokenizing step, download this file and put it into the data section.