View Code? Open in Web Editor
NEW
License: Apache License 2.0
Python 99.19%
Shell 0.81%
regular_gpt's Introduction
Implementation of Transformer Working Memory Enables Regular Language Reasoning and Natural Language Length Extrapolation
- This repository is modified from the codebase of Neural Networks and the Chomsky Hierarchy. Please check README_original.md for installation details.
- Please check neural_networks_chomsky_hierarchy/run.sh for the hyperparameters we used for our regular language experiments.
regular_gpt's People
Watchers