Document Context Language Models (DCLMs)
A set of document-level language models with contextual information. Please refer to our ICLR 2016 submission for more technical details.
You need the Boost C++ libraries (>=1.56) to save/load word vocabulary and trained models.
First you need to fetch the cnn library into the same folder, then follow the instruction to get additional libraries and compile cnn.
To compile all DCLMs, run
It will produce three executable files: main-dclm, dam, and baseline.
- baseline refers to the RNNLM model with contextual information
- dam refers to the attentional DCLM
- main-dclm includes other four models: context-to-output DCLM, context-to-context DCLM, RNNLM without sentence boundary, and Hierarchical RNNLM.