Model validation will fail for CB cases

VowpalWabbit / reinforcement_learning

Interaction-side integration library for Reinforcement Learning loops: Predict, Log, [Learn,] Update

MIT License

75 stars 40 forks source link

Model validation will fail for CB cases #124

Open slahabar opened 4 years ago

slahabar commented 4 years ago

In RLClientLib/vw_model/vw_model.cc the default value for MODEL_VW_INITIAL_COMMAND_LINE is changed to CCB. The default behavior has now changed for cb. The compatibility check for APS model s will always fail.

Change the default command line to cb.

jackgerrits commented 4 years ago

When there is no model for CB, it does not use the safe_vw factory. It instead uses an explicit explore only mode. See here: https://github.com/VowpalWabbit/reinforcement_learning/blob/bf4d7bb54b79703831b46af74edbd7047d4fa5af/rlclientlib/live_model_impl.cc#L64