search

raghakot / keras-vis

Neural network visualization toolkit for keras
https://raghakot.github.io/keras-vis
MIT License
2.98k stars 664 forks source link

Cannot convert a symbolic Keras input/output to a numpy array... #234

Open abrarum opened 3 years ago

abrarum commented 3 years ago

Tensorflow version: 2.4.0

Issue: I tried to reproduce this example code but results in an error (it works in Tensorflow version 2.2.0): "TypeError: Cannot convert a symbolic Keras input/output to a numpy array. This error may indicate that you're trying to pass a symbolic value to a NumPy call, which is not supported. Or, you may be trying to pass Keras symbolic inputs/outputs to a TF API that does not register dispatching, preventing Keras from automatically converting the API call to a lambda layer in the Functional Model."

Trace:

TypeError Traceback (most recent call last)

in 15 #print(x_test[idx]) 16 ---> 17 grads = visualize_saliency(model, layer_idx, filter_indices=class_idx, seed_input=x_test[idx]) 18 # Plot with 'jet' colormap to visualize as a heatmap. 19 plt.imshow(grads, cmap='jet') ~/Abrar/miniconda3/lib/python3.8/site-packages/vis/visualization/saliency.py in visualize_saliency(model, layer_idx, filter_indices, seed_input, backprop_modifier, grad_modifier) 124 (ActivationMaximization(model.layers[layer_idx], filter_indices), -1) 125 ] --> 126 return visualize_saliency_with_losses(model.input, losses, seed_input, grad_modifier) 127 128 ~/Abrar/miniconda3/lib/python3.8/site-packages/vis/visualization/saliency.py in visualize_saliency_with_losses(input_tensor, losses, seed_input, grad_modifier) 71 weighted `losses`. 72 """ ---> 73 opt = Optimizer(input_tensor, losses, norm_grads=False) 74 grads = opt.minimize(seed_input=seed_input, max_iter=1, grad_modifier=grad_modifier, verbose=False)[1] 75 ~/Abrar/miniconda3/lib/python3.8/site-packages/vis/optimizer.py in __init__(self, input_tensor, losses, input_range, wrt_tensor, norm_grads) 50 51 # Compute gradient of overall with respect to `wrt` tensor. ---> 52 grads = K.gradients(overall_loss, self.wrt_tensor)[0] 53 if norm_grads: 54 grads = grads / (K.sqrt(K.mean(K.square(grads))) + K.epsilon()) ~/Abrar/miniconda3/lib/python3.8/site-packages/keras/backend/tensorflow_backend.py in symbolic_fn_wrapper(*args, **kwargs) 73 if _SYMBOLIC_SCOPE.value: 74 with get_graph().as_default(): ---> 75 return func(*args, **kwargs) 76 else: 77 return func(*args, **kwargs) ~/Abrar/miniconda3/lib/python3.8/site-packages/keras/backend/tensorflow_backend.py in gradients(loss, variables) 3023 if _is_tf_1(): 3024 return tf.gradients(loss, variables, colocate_gradients_with_ops=True) -> 3025 return tf.gradients(loss, variables) 3026 3027 ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/ops/gradients_impl.py in gradients_v2(ys, xs, grad_ys, name, gate_gradients, aggregation_method, stop_gradients, unconnected_gradients) 313 # pylint: disable=protected-access 314 with ops.get_default_graph()._mutation_lock(): --> 315 return gradients_util._GradientsHelper( 316 ys, xs, grad_ys, name, True, gate_gradients, 317 aggregation_method, stop_gradients, ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/ops/gradients_util.py in _GradientsHelper(ys, xs, grad_ys, name, colocate_gradients_with_ops, gate_gradients, aggregation_method, stop_gradients, unconnected_gradients, src_graph) 526 # cluster ops for compilation. 527 gradient_uid = ops.get_default_graph().unique_name("uid") --> 528 ys = ops.convert_n_to_tensor_or_indexed_slices(ys, name="y") 529 xs = [ 530 x.handle if resource_variable_ops.is_resource_variable(x) else x ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/indexed_slices.py in convert_n_to_tensor_or_indexed_slices(values, dtype, name) 383 value. 384 """ --> 385 return internal_convert_n_to_tensor_or_indexed_slices( 386 values=values, dtype=dtype, name=name, as_ref=False) 387 ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/indexed_slices.py in internal_convert_n_to_tensor_or_indexed_slices(values, dtype, name, as_ref) 355 n = None if name is None else "%s_%d" % (name, i) 356 ret.append( --> 357 internal_convert_to_tensor_or_indexed_slices( 358 value, dtype=dtype, name=n, as_ref=as_ref)) 359 return ret ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/indexed_slices.py in internal_convert_to_tensor_or_indexed_slices(value, dtype, name, as_ref) 316 return value 317 else: --> 318 return ops.convert_to_tensor(value, dtype=dtype, name=name, as_ref=as_ref) 319 320 ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/profiler/trace.py in wrapped(*args, **kwargs) 161 with Trace(trace_name, **trace_kwargs): 162 return func(*args, **kwargs) --> 163 return func(*args, **kwargs) 164 165 return wrapped ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/ops.py in convert_to_tensor(value, dtype, name, as_ref, preferred_dtype, dtype_hint, ctx, accepted_result_types) 1538 1539 if ret is None: -> 1540 ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref) 1541 1542 if ret is NotImplemented: ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/constant_op.py in _constant_tensor_conversion_function(v, dtype, name, as_ref) 337 as_ref=False): 338 _ = as_ref --> 339 return constant(v, dtype=dtype, name=name) 340 341 ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/constant_op.py in constant(value, dtype, shape, name) 262 ValueError: if called on a symbolic tensor. 263 """ --> 264 return _constant_impl(value, dtype, shape, name, verify_shape=False, 265 allow_broadcast=True) 266 ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/constant_op.py in _constant_impl(value, dtype, shape, name, verify_shape, allow_broadcast) 279 tensor_value = attr_value_pb2.AttrValue() 280 tensor_value.tensor.CopyFrom( --> 281 tensor_util.make_tensor_proto( 282 value, dtype=dtype, shape=shape, verify_shape=verify_shape, 283 allow_broadcast=allow_broadcast)) ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/framework/tensor_util.py in make_tensor_proto(values, dtype, shape, verify_shape, allow_broadcast) 433 434 if _is_array_like(values): --> 435 values = np.asarray(values) 436 437 # We first convert value to a numpy array or scalar. ~/Abrar/miniconda3/lib/python3.8/site-packages/numpy/core/_asarray.py in asarray(a, dtype, order) 81 82 """ ---> 83 return array(a, dtype, copy=False, order=order) 84 85 ~/Abrar/miniconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/keras_tensor.py in __array__(self) 271 272 def __array__(self): --> 273 raise TypeError( 274 'Cannot convert a symbolic Keras input/output to a numpy array. ' 275 'This error may indicate that you\'re trying to pass a symbolic value ' TypeError: Cannot convert a symbolic Keras input/output to a numpy array. This error may indicate that you're trying to pass a symbolic value to a NumPy call, which is not supported. Or, you may be trying to pass Keras symbolic inputs/outputs to a TF API that does not register dispatching, preventing Keras from automatically converting the API call to a lambda layer in the Functional Model.
Sciprios commented 3 years ago

I have been getting the same issue, I have tried upgrading tensorflow, keras and the keras-vis packages to no avail.