hls_path = 'denoise_mnist'

config = hls4ml.utils.config_from_keras_model(model, granularity='model')

config['Model'] = {}

config['Model']['ReuseFactor'] = 64

config['Model']['Strategy'] = 'Resource'

config['Model']['Precision'] = 'ap_fixed<32,16>'

hls_model = hls4ml.converters.convert_from_keras_model(model,
                                                       hls_config=config,
                                                       output_dir=hls_path,
                                                       backend='VivadoAccelerator',
                                                       io_type='io_stream',
                                                       board='pynq-z2')