Home

Welcome to the MachineRC wiki!

Progress Report

• Wrote data parser for SQuad.
• Made baseline to (almost) work.
• Figured how to used pretrained embeddings for the project
• Figured out the nuts and bolts of RNN networks by using different kinds of RNNs to do all sorts of things like regression, sentence classification, MNIST image classification. The code's here https://github.com/wasimusu/RNNs.
• Figured out training models on Google Cloud Platform. It has its' own nuisance.

TODO

• Figure out the architecture details from the paper that we're trying to replicate.
• Figure out how to speed up training on GCP.

More things to sort out

• Vocabulary size
• How to deal with out-of-vocabulary words