Error when Converting Quantized ONNX Model to HLS with hls4ml: Unsupported Operation Type Quant

[dianamahdi]

Hello,

I'm working on converting a quantized ONNX model to HLS for deployment on an FPGA using hls4ml. However, I encounter an error related to unsupported operation types during the conversion process. Below is the relevant code snippet and the error message:

import hls4ml
model_onnx = "cybsec-mlp-ready.onnx"
hls_config = hls4ml.utils.config_from_onnx_model(model_onnx, granularity='name')

cfg = hls4ml.converters.create_config(backend='Vivado')
cfg['ClockPeriod'] = 10
cfg['HLSConfig'] = hls_config
cfg['OnnxModel'] = model_onnx
cfg['Backend'] = 'VivadoAccelerator'
cfg['OutputDir'] = 'qdnn-dsd_accelerator/'
cfg['Board'] = 'pynq-z2'

model_hls = hls4ml.converters.onnx_to_hls(cfg)
model_hls.compile()

Interpreting Model ...
Output layers: ['/8/Softmax']
Input shape: [None, 16]
Topology:

Exception Traceback (most recent call last)
Cell In[3], line 13
10 cfg['OutputDir'] = 'qdnn-dsd_accelerator/'
11 cfg['Board'] = 'pynq-z2'
---> 13 model_hls = hls4ml.converters.onnx_to_hls(cfg)
15 model_hls.compile()

File ~\AppData\Local\Programs\Python\Python311\Lib\site-packages\hls4ml\converters\onnx_to_hls.py:281, in onnx_to_hls(config)
279 for node in graph.node:
280 if node.op_type not in supported_layers:
--> 281 raise Exception(f'ERROR: Unsupported operation type: {node.op_type}')
283 # If not the first layer then input shape is taken from last layer's output
284 if layer_counter != 0:

Exception: ERROR: Unsupported operation type: Quant

[vloncar]

Try #832