ValueError: Validation Failed

[RhinoInani]

Hello,
I am running into an issue where the auto converter keeps failing due to validation in the model. I understand the basics of why this is occurring but am not sure about the in-depth detail. Could someone please provide some more details about how I can fix this issue and/or why it could be occurring.

CODE:

feed_dict_list = []

for batch in tqdm(val_loader_onnx, desc="Processing Batches"):
    inputs, labels = batch
    inputs = inputs.numpy()
    labels = labels.numpy()

    feed_dict = {
        'input': inputs,
    }
    feed_dict_list.append(feed_dict)


model_fp16 = auto_convert_mixed_precision(model, feed_dict, rtol=0.01, atol=0.001, keep_io_types=True)

ERROR:

Traceback (most recent call last):
  File "onnx_converter.py", line 73, in <module>
    model_fp16 = auto_convert_mixed_precision(model, feed_dict, rtol=0.01, atol=0.001)
  File "****/miniconda3/envs/internimage/lib/python3.7/site-packages/onnxconverter_common/auto_mixed_precision.py", line 81, in auto_convert_mixed_precision
    raise ValueError("validation failed for model with all nodes in node_block_list")
ValueError: validation failed for model with all nodes in node_block_list

PS:
Another problem I was running into with the autoconverter was when I was running feed_dict_list in the line of code above to run the auto_convert_mixed_precision it doesn't run fully and doesn't generate a new onnx file and just quits with no warning or error. This occurs when keep_io_types is True or False. Is there any reason for this? Could someone please explain this to me.

This is an example of the end of the output from when I use feed_dict_list instead of feed_dict:

[-1.0549891 , -1.1072767 , -1.1421349 , ...,  0.2870589 ,
           0.2870589 ,  0.32191727]]]], dtype=float32)}], None

Thanks,
Rhino