RTLM

Introduction

This repository contains the code for the paper "Transformer Language Models with LSTM-based Cross-utterance Information Representation". The code is mainly adapted from the Transformer XL PyTorch implementation. Single GPU version is implemented.

Prerequisite

PyTorch 1.0.0

Training

To train on AMI text data
bash run_AMI.sh train --work_dir PATH_TO_WORK_DIR

To train on SWBD text data
bash run_SWBD.sh train --work_dir PATH_TO_WORK_DIR

N-best Rescoring

Rescoring AMI nbest list bash AMInbest.sh
Rescoring SWB nbest list bash SWBDnbest.sh
Rescoring RT03 nbest list bash RTnbest.sh

Note that the path to the trained LM (--model) and to the nbest list (--nbest) should be modified to your own path.