These are 5 example models; in order of complexity
- app_no_flash - a single model, no flash memory used. This is the fastest but most pressure on internal memory.
- app_flash_single_model - a single model, with learned parameters in flash memory. This removes a lot of pressure on internal memory.
- app_flash_two_models - two models, with learned parameters in flash memory.
- app_flash_two_models_one_arena - two models, with learned parameters in flash memory. The models share a single tensor arena (scratch memory).