4.3.0 yolov4 int4量化onnx报错

已反馈内部

```
Traceback (most recent call last):
  File "/root/deepsparkinference/models/cv/object_detection/yolov4/ixrt/.//quant.py", line 49, in <module>
    static_quantize(args.model,
  File "/usr/local/corex-4.3.0/lib64/python3/dist-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/api/quantization.py", line 249, in static_quantize
    ret = pipline.run()[0]
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/api/pipeline.py", line 57, in run
    graph = comp(graph)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/quantizer/algorithm/base_quantizer.py", line 332, in __call__
    self.finetune()
  File "/usr/local/corex-4.3.0/lib64/python3/dist-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/quantizer/algorithm/static_quantizer.py", line 36, in finetune
    return run_graph_one_epoch(
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/quantizer/utils.py", line 59, in run_graph_one_epoch
    executor.execute_graph(graph, inputs)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/ir/base_executor.py", line 73, in execute_graph
    return self._execute_graph(graph, inputs)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/ir/base_executor.py", line 85, in _execute_graph
    output = self.exec_operator(graph, op)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/ir/base_executor.py", line 94, in exec_operator
    return self.exec_nonquant_operator(graph, op)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/ir/base_executor.py", line 100, in exec_nonquant_operator
    return op_func(self, op, inputs, attr)
  File "/usr/local/lib/python3.10/site-packages/ixrt/deploy/backend/torch/executor/operators/_operators.py", line 241, in call_conv
    return F.conv2d(**conv_params)
RuntimeError: Input type (unsigned char) and bias type (float) should be the same
```

DeepSpark/DeepSparkInference

Content Risk Flag

Comments (0)

DeepSpark/DeepSparkInference .gitee-modal { width: 500px !important; }

Content Risk Flag