Reproduce “Learning End-to-End Goal-Oriented Dialog” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Antoine Bordes, Y-Lan Boureau, Jason Weston
The paper tests whether a Memory Network can handle restaurant-booking dialogs by collecting missing details, calling a knowledge base, and using returned facts. It evaluates candidate ranking across simulated tasks and converted real-dialog data, not free-form generation. On full dialogs, the Memory Network without match-type features reaches 96.1% per-response but 49.4% per-dialog accuracy, showing uneven performance rather than reliable booking completion.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.