[Tensorflow] IndexError: list index (0) out of range

[peiwenhuang27]

Environment: Google Colaboratory
Tensorflow (official): 2.6.0
Neural Compressor: 1.7.2

Hi, I was quantizing a pre-trained Keras model (also in Tensorflow 2.6) with post training quantization method when the following error occurred:

2021-10-28 08:33:25 [INFO] Pass Quantization elapsed time: 9205.52 ms
2021-10-28 08:33:42 [INFO] Pass QuantizedRNNConverter elapsed time: 300.62 ms
Traceback (most recent call last):
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tensorflow.py", line 462, in quantize
    data_loader=data_loader).convert()
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tf_utils/graph_converter.py", line 249, in convert
    post_cse_graph_def = PostCseOptimizer(model.graph_def).do_transformation()
AttributeError: 'NoneType' object has no attribute 'graph_def'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tf_utils/graph_converter.py", line 571, in quantize
    self._fuse_requantize_with_fused_quantized_node()
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tf_utils/graph_converter.py", line 709, in _fuse_requantize_with_fused_quantized_node
    self._tmp_graph_def).do_transformation()
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tf_utils/graph_rewriter/int8/fuse_matmul_requantize.py", line 108, in do_transformation
    min_input_value = (min_input_node.attr['value'].tensor.float_val)[0]
IndexError: list index (0) out of range
2021-10-28 08:33:43 [ERROR] Fail to quantize graph due to list index (0) out of range.
2021-10-28 08:33:43 [ERROR] Unexpected exception AttributeError("'NoneType' object has no attribute 'graph_def'") happened during tuning.
Traceback (most recent call last):
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tensorflow.py", line 462, in quantize
    data_loader=data_loader).convert()
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tf_utils/graph_converter.py", line 249, in convert
    post_cse_graph_def = PostCseOptimizer(model.graph_def).do_transformation()
AttributeError: 'NoneType' object has no attribute 'graph_def'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/experimental/quantization.py", line 151, in execute
    self.strategy.traverse()
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/strategy/strategy.py", line 333, in traverse
    tune_cfg, self.model, self.calib_dataloader, self.q_func)
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/utils/utility.py", line 240, in fi
    res = func(*args, **kwargs)
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tensorflow.py", line 475, in quantize
    data_loader=data_loader).convert()
  File "/usr/local/lib/python3.7/dist-packages/neural_compressor/adaptor/tf_utils/graph_converter.py", line 249, in convert
    post_cse_graph_def = PostCseOptimizer(model.graph_def).do_transformation()
AttributeError: 'NoneType' object has no attribute 'graph_def'
2021-10-28 08:33:43 [ERROR] Specified timeout or max trials is reached! Not found any quantized model which meet accuracy goal. Exit.
Traceback (most recent call last):
  File "main.py", line 115, in <module>
    evaluate_opt_graph.run()
  File "main.py", line 100, in run
    q_model.save(self.args.output_graph) 
AttributeError: 'NoneType' object has no attribute 'save'

The complete log is attached below.
error_log.txt

If helpful, please find the quantization config, pre-trained model and script here (please request access via email)

Thanks!

[peiwenhuang27]

I think this may be a duplicate of an issue I posted earlier #26 , should I close this and update info with comment in the previous issue? Thanks!

[guomingz]

@peiwenhuang27 i applied the access for script this morning but i didn't receive any feedback.

Would u please send it to me directly？

[peiwenhuang27]

@guomingz Sorry I didn't see the request at first! I did approve the request later, though, perhaps there wasn't notification email sent to you. Could you click the link again with your email address to check if you have access now? Thanks!

[hshen14]

@peiwenhuang27 Sorry for keeping this for a long time. May I know whether the issue is still there in latest version?

[chensuyue]

Please re-open the issue if detected again.