def test_model_saving_after_training(self):
init_directories()
model_name = "model_1"
model = build_model(model_name)
self.assertEqual(model.name, 'model_1')
board, player = game_init()
policies, values = model.predict(board)
try:
os.remove('test.h5')
except:
pass
model.save('test.h5')
self_play(model, n_games=2, mcts_simulations=32)
train(model, game_model_name=model.name, epochs=2)
self.assertEqual(model.name, 'model_2')
policies2, values2 = model.predict(board)
self.assertFalse(np.array_equal(values, values2))
self.assertFalse(np.array_equal(policies, policies2))
model3 = load_model('test.h5', custom_objects={'loss': loss})
policies3, values3 = model3.predict(board)
self.assertTrue(np.array_equal(values, values3))
self.assertTrue(np.array_equal(policies, policies3))
os.remove('test.h5')
def setUp(self):
init_directories()
model_name = "model_1"
model = create_initial_model(name=model_name)
best_model = load_best_model()
if best_model.name == model.name:
train(model, game_model_name=best_model.name)
evaluate(best_model, model)
# We save wether or not it was a better model
full_filename = os.path.join(conf['MODEL_DIR'], conf['BEST_MODEL'])
model.save(full_filename)
else:
model = best_model
self.model = model
def test_model_saving(self):
init_directories()
model_name = "model_1"
model = build_model(model_name)
board, player = game_init()
policies, values = model.predict(board)
try:
os.remove('test.h5')
except:
pass
model.save('test.h5')
model2 = load_model('test.h5', custom_objects={'loss': loss})
policies2, values2 = model2.predict(board)
self.assertTrue(np.array_equal(values, values2))
self.assertTrue(np.array_equal(policies, policies2))
os.remove('test.h5')
def setUp(self):
init_directories()
model_name = "model_1"
best_model = load_best_model()
self.model = best_model
self.board, player = game_init()
