probcomp / GenExperimental.jl

Featherweight embedded probabilistic programming language and compositional inference programming library
MIT License
17 stars 2 forks source link

Goal Inference: Load up weights for Neural Network #41

Closed mirisr closed 7 years ago

mirisr commented 7 years ago

Although the user experience should wait the hour or so to train the network, it would be handy for people testing the notebook like Vikash and myself to have the network trained beforehand.

marcoct commented 7 years ago

I have a version of the network weights saved in trained_waypoint_predictor_parameters.json in https://github.com/probcomp/gen-examples/commit/a2226f8875e2ea98197ec3c50309832dda051cc2.