Create trainer.py scaffolding to ease training of model architecture (taken from model.py)

alikureishy / Automata-Capstone

Capstone repository (pulled from /verichev)

MIT License

2 stars 0 forks source link

Create trainer.py scaffolding to ease training of model architecture (taken from model.py) #4

Closed alikureishy closed 5 years ago

alikureishy commented 5 years ago

New params: /tl_classifier/training_data_folder = <...> /tl_classifier/model_folder = <...>

Location: ros/src/tl_detector/light_classification/trainer.py

alikureishy commented 5 years ago

The training model can be built using the "create_model_for_training()" method in model.py:

from model import create_model_for_training input, output, loss_function, optimizer = create_model_for_training(....) ... ...# Wake up when training request received ...# Read training data ...# Train model ...# Write trained model to disk