The paper says it also need a number of labeled SQL to train?

crazydonkey200 / neural-symbolic-machines

Neural Symbolic Machines is a framework to integrate neural networks and symbolic representations using reinforcement learning, with applications in program synthesis and semantic parsing.

Apache License 2.0

377 stars 69 forks source link

The paper says it also need a number of labeled SQL to train? #24

Closed guotong1988 closed 5 years ago

guotong1988 commented 5 years ago

Thank you very much.

guotong1988 commented 5 years ago

Am I right?

crazydonkey200 commented 5 years ago

Hi, the model doesn't need labeled SQL to train. It only uses question and answer pairs as weak supervision. Please see Section 4 for more details.

guotong1988 commented 5 years ago

Thank you！

guotong1988 commented 5 years ago

What does load_programs in experiment.py do? @crazydonkey200 Thank you!

crazydonkey200 commented 5 years ago

It refers to programs found from random search, see section 5.1 training details.