- featurize
During training frozen vertices/layers can be treated as "featurizing" the input
The forward pass th
- <init>
Will modify the given MLN (in place!) to freeze layers (hold params constant
during training) specif
- unfrozenGraph
Returns the unfrozen subset of the original computation graph as a computation
graph Note that with
- copyOrigParamsToSubsetGraph
- copyParamsFromSubsetGraphToOrig
- copyParamsFromSubsetMLNToOrig
- initHelperGraph
Runs through the comp graph and saves off a new model that is simply the
"unfrozen" part of the orig
- initHelperMLN