Latent Relation Language Models

Authors: Hiroaki Hayashi, Zecong Hu, Chenyan Xiong, Graham Neubig7911-7918

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments demonstrate empirical improvements over both word-based language models and a previous approach that incorporates knowledge graph information.
Researcher Affiliation Collaboration 1Carnegie Mellon University, 2Microsoft Research AI {hiroakih, zeconghu, gneubig}@cs.cmu.edu, Chenyan.Xiong@microsoft.com
Pseudocode Yes Algorithm 1 Generative Process of LRLM
Open Source Code Yes Equal Contribution. Code & Data: https://github.com/neulab/lrlm.
Open Datasets Yes Wiki Facts (Ahn et al. 2016) is a collection of Wikipedia articles restricted to /film/actor domain entities in Freebase (Bollacker et al. 2008).
Dataset Splits Yes Since official splits for evaluation are not provided, we follow previous work and performed a random split of 80/10/10%.
Hardware Specification No The paper mentions implementing models in Py Torch and training them, but does not provide specific details on the hardware used for experiments, such as GPU or CPU models.
Software Dependencies No We implement all models in Py Torch (Paszke et al. 2017).
Experiment Setup Yes Training details and hyperparameters are summarized in Appendix B.