search

tensorflow / addons

Useful extra functionality for TensorFlow 2.x maintained by SIG-addons
Apache License 2.0
1.69k stars 610 forks source link

AttributeError: 'ResourceVariable' object has no attribute 'overwrite_with_gradient' #2871

Open Maqingkuii opened 2 months ago

Maqingkuii commented 2 months ago

System information

Describe the bug Compiling model... 'compile' took 0.007720 s

Training model...

Step Train loss Test loss Test metric 0 [9.10e-03, 7.54e-03, 2.72e-02, 1.26e+00, 2.77e-01] [9.10e-03, 7.54e-03, 2.72e-02, 1.26e+00, 2.77e-01] []
Traceback (most recent call last): File "E:\python\Paper_Programs\05_deepxde\examples\pinn_inverse\Navier_Stokes_inverse.py", line 124, in loss_history, train_state = model.train( File "E:\python\Paper_Programs\05_deepxde\deepxde\utils\internal.py", line 22, in wrapper result = f(*args, **kwargs) File "E:\python\Paper_Programs\05_deepxde\deepxde\model.py", line 650, in train self._train_sgd(iterations, display_every) File "E:\python\Paper_Programs\05_deepxde\deepxde\model.py", line 668, in _train_sgd self._train_step( File "E:\python\Paper_Programs\05_deepxde\deepxde\model.py", line 560, in _train_step self.train_step(inputs, targets, auxiliary_vars) File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\tensorflow\python\util\traceback_utils.py", line 153, in error_handler raise e.with_traceback(filtered_tb) from None File "C:\Users\86186\AppData\Local\Temp__autograph_generated_file66hkpahi.py", line 16, in tftrain_step ag__.converted_call(ag.ld(opt).apply_gradients, (ag__.converted_call(ag.ld(zip), (ag.ld(grads), ag__.ld(trainable_variables)), None, fscope),), None, fscope) File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\keras\src\optimizers\base_optimizer.py", line 282, in apply_gradients self.apply(grads, trainable_variables) File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\keras\src\optimizers\base_optimizer.py", line 329, in apply self._overwrite_variables_directly_with_gradients( File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\keras\src\optimizers\base_optimizer.py", line 618, in _overwrite_variables_directly_with_gradients if v.overwrite_with_gradient: AttributeError: in user code:

File "E:\python\Paper_Programs\05_deepxde\deepxde\model.py", line 249, in train_step  *
    opt.apply_gradients(zip(grads, trainable_variables))
File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\keras\src\optimizers\base_optimizer.py", line 282, in apply_gradients  **
    self.apply(grads, trainable_variables)
File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\keras\src\optimizers\base_optimizer.py", line 329, in apply
    self._overwrite_variables_directly_with_gradients(
File "C:\Users\86186\anaconda\envs\05_deepxde\lib\site-packages\keras\src\optimizers\base_optimizer.py", line 618, in _overwrite_variables_directly_with_gradients
    if v.overwrite_with_gradient:

AttributeError: 'ResourceVariable' object has no attribute 'overwrite_with_gradient'

Can anyone solve this problem?

arezayan commented 2 months ago

I encountered the same issue.