Changed the way we store summary items. Previously they were stored in the model_entity. Now they
are stored in a new kind of entity: model_summary_items_entity. This prevents us from exceeding the
limits of a single entity.
In the models list, increase the time before retrieving a model entity during training to 5 minutes.
This reduces the cost of app engine requests during training.
Added code to use GPUs. So far it doesn't not produce usable models, so it is stilled turned off.
Changed the way we store summary items. Previously they were stored in the model_entity. Now they are stored in a new kind of entity: model_summary_items_entity. This prevents us from exceeding the limits of a single entity.
In the models list, increase the time before retrieving a model entity during training to 5 minutes. This reduces the cost of app engine requests during training.
Added code to use GPUs. So far it doesn't not produce usable models, so it is stilled turned off.