Andrew Hundt

Doing a 300 epoch training run as of https://github.com/cpaxton/costar_plan/pull/410 b9a998b62faaec0f94444842943558755a3baced with both the pixel-wise and single prediction levine models

The run is complete, the results look pretty good for the single prediction model! There was noticeable overfitting by epoch 30 so I stopped then. The peak performance model on...

Note: The first run of the pixel-wise model with a single pixel loss did not make any training progress for the first 5 epochs with validation results equivalent to random,...

This visualization enhancement was suggested by @cpaxton, could you add your thoughts on this issue description?

Initial work in https://github.com/cpaxton/costar_plan/pull/429, this issue is still in progress.

Good progress towards this in https://github.com/cpaxton/costar_plan/pull/437

progress has been made in https://github.com/cpaxton/costar_plan/pull/410

Almost there! we might want to ask the following questions: 1. can cells be created separately fairly easily with the current code? 2. Can fashion mnist be trained well with...

Here is the pull request with this work in progress: https://github.com/cpaxton/costar_plan/pull/520

There is some work on this in both tensorflow and keras-contrib: https://github.com/tensorflow/tensorflow/pull/12549