Hi, I am trying to convert my model to TFlite and getting troubles because of F.interpolate? some how it has control flow in it and so TFlite converter dies. I also tried equivalent repeat_interleave it works fine but pytorch refuses to convert that n ONNX model. Could you please offer me some analog of F.interpolate based on very basic pytorch functions, speed is actually has secondary priority.
Do you get an error message when the converter dies?
If so, could you post it please?
Hi @ptrblck!
I am converting this to ONNX then to *.pb (GraphDef using onnx-tensorflow project, which is OK in execution) and then to TFlite which converter dies.
I am using CUDA_VISIBLE_DEVICES=-1 and #613 fix for onnx_tensorflow project (for Dilation2d issue)
ONNX : 1.6.0
TF: 2.1.0
troch: 1.5.0
class SimpleNet(torch.nn.Module):
def __init__(self):
super(SimpleNet, self).__init__()
def forward(self, x):
x = F.interpolate(x, scale_factor=2, mode='nearest')#x.view(1, 3, 672, 672)
return x
Most interesting part of output:
2020-05-13 10:47:51.019359: I tensorflow/lite/toco/import_tensorflow.cc:659] Converting unsupported operation: If
2020-05-13 10:47:51.019406: F tensorflow/lite/toco/import_tensorflow.cc:114] Check failed: attr.value_case() == AttrValue::kType (1 vs. 6)
Fatal Python error: Aborted
Whole output:
tensorflow.lite.python.convert.ConverterError: See console for info.
2020-05-13 10:47:50.562338: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer.so.6'; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/marat/anaconda3/lib
2020-05-13 10:47:50.562423: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer_plugin.so.6'; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda/lib64:/home/marat/anaconda3/lib
2020-05-13 10:47:50.562430: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
2020-05-13 10:47:51.003834: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2020-05-13 10:47:51.009672: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3407930000 Hz
2020-05-13 10:47:51.010213: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55a180ac8c00 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-05-13 10:47:51.010228: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version
2020-05-13 10:47:51.011847: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-05-13 10:47:51.014197: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected
2020-05-13 10:47:51.014232: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: saturn
2020-05-13 10:47:51.014237: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: saturn
2020-05-13 10:47:51.014283: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:200] libcuda reported version is: 440.64.0
2020-05-13 10:47:51.014316: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:204] kernel reported version is: 440.64.0
2020-05-13 10:47:51.014321: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:310] kernel version seems to match DSO: 440.64.0
2020-05-13 10:47:51.019359: I tensorflow/lite/toco/import_tensorflow.cc:659] Converting unsupported operation: If
2020-05-13 10:47:51.019406: F tensorflow/lite/toco/import_tensorflow.cc:114] Check failed: attr.value_case() == AttrValue::kType (1 vs. 6)
Fatal Python error: Aborted
Current thread 0x00007f68c1c05700 (most recent call first):
File "/home/marat/anaconda3/envs/cexp/lib/python3.7/site-packages/tensorflow_core/lite/toco/python/toco_from_protos.py", line 56 in execute
File "/home/marat/anaconda3/envs/cexp/lib/python3.7/site-packages/absl/app.py", line 250 in _run_main
File "/home/marat/anaconda3/envs/cexp/lib/python3.7/site-packages/absl/app.py", line 299 in run
File "/home/marat/anaconda3/envs/cexp/lib/python3.7/site-packages/tensorflow_core/python/platform/app.py", line 40 in run
File "/home/marat/anaconda3/envs/cexp/lib/python3.7/site-packages/tensorflow_core/lite/toco/python/toco_from_protos.py", line 93 in main
File "/home/marat/anaconda3/envs/cexp/bin/toco_from_protos", line 8 in <module>
Aborted (core dumped)
Thanks for the update. This seems to be a TensorFlow issue, and I’m not experienced enough with this framework, so you might get a better answer of StackOverflow.