Closed pigking0126 closed 9 months ago
I am also having the same problem please help me
Hey, thanks.
Usually, you can do the conversion in the following way:
tf2onnx.convert.from_keras(model, output_path=onnx_model_path)
But, I'll create a change, so if you don't have installed tf2onnx and onnx packages, it will raise an error when you import Model2onnx callback
I don't want to add these packages into mltu, because if we use PyTroch for training, they are not necessary. PyTorch have their own conversion integrated into torch
What should I do to fix that error?
I tried installing tf2onnx and onnx by pip install but when train reached epoch 317 the process stopped, here are the error details please help me check
Epoch 317: early stopping 2023-09-13 13:24:46.914593: I tensorflow/core/grappler/devices.cc:75] Number of eligible GPUs (core count >= 8, compute capability >= 0.0): 0 (Note: TensorFlow was not compiled with CUDA or ROCm support) 2023-09-13 13:24:46.931476: I tensorflow/core/grappler/clusters/single_machine.cc:357] Starting new session 2023-09-13 13:24:49.225700: I tensorflow/core/grappler/devices.cc:75] Number of eligible GPUs (core count >= 8, compute capability >= 0.0): 0 (Note: TensorFlow was not compiled with CUDA or ROCm support) 2023-09-13 13:24:49.242155: I tensorflow/core/grappler/clusters/single_machine.cc:357] Starting new session 2023-09-13 13:24:50.268365: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:375] MLIR V1 optimization pass is not enabled
@pythonlessons
What gpu you have? this error is not related with code, the issue might be with drivers or tensorflow, not sure...
This is my gpu, can I run it on cpu?
Yes, you gpu is too old, I think it doesn't support onnx, you can try convert it using cpu
In new version 1.1.1
the Import error will be raised then you try to:
from mltu.tensorflow.callbacks import Model2onnx
if you don't have tf2onnx and onnx installed
it seems that using 'pip install mltu' doesn't install tf2onnx
cause after the train process has been done, it will return error: No module named tf2onnx
so now i only have model.h5, but since I use CTCloss and CWERMetric and stuff like this, I have a hard time trying to export it to onnx file is there any function in this module that can do this for me? I can only find callback function not direct tf to onnx.