“Neural Networks for Modeling Source Code Edits” https://arxiv.org/abs/1904.02818
Seems like a fascinating line of inquiry, though possibly problematic from the perspective of unaligned AI self-improvement.
Following on that:
“Mathematical Reasoning Abilities of Neural Models,” https://arxiv.org/pdf/1904.01557.pdf
They have proposed a procedurally-generated data set for testing whether a model is capable of the same types of mathematical reasoning as humans.
“Neural Networks for Modeling Source Code Edits” https://arxiv.org/abs/1904.02818
Seems like a fascinating line of inquiry, though possibly problematic from the perspective of unaligned AI self-improvement.
Following on that:
“Mathematical Reasoning Abilities of Neural Models,” https://arxiv.org/pdf/1904.01557.pdf
They have proposed a procedurally-generated data set for testing whether a model is capable of the same types of mathematical reasoning as humans.