def train_mnist_worker(params):
net_id = params.get('net-id', 'nn')
layers = [784]
layers.extend([int(i) for i in params.get('layers', [15])])
layers.append(10)
net_params = {}
net_params['epochs'] = int(params.get('epochs', 1))
net_params['mini_batch_size'] = int(params.get('mini-batch-size', 4))
net_params['eta'] = float(params.get('eta', 0.1))
net_params['lmbda'] = float(params.get('lmbda', 0.0001))
net_params['layers'] = layers
redis.set(redis_key('params', net_id), json.dumps(net_params))
redis.set(redis_key('status', net_id), 'train_mnist: started')
net = Network(layers)
training_data, validation_data, test_data = load_data_wrapper()
redis.set(redis_key('status', net_id), 'train_mnist: training with mnist data')
net.SGD(training_data, net_params['epochs'],
net_params['mini_batch_size'],
net_params['eta'],
net_params['lmbda'])
redis.set(redis_key('data', net_id), net.tostring())
redis.set(redis_key('status', net_id), 'train_mnist: trained')
