> 1848 cancellation_manager=cancellation_manager) 1849 1850 def _call_flat(self, args, captured_inputs, Resource_variable_ops.BaseResourceVariable))], 1847 > 2829 return graph_function._filtered_call(args, kwargs) # pylint: disable=protected-access 2830 2831 _filtered_call(self, args, kwargs, cancellation_manager) 1846 Self._maybe_define_function(args, kwargs) In call(self, *args, **kwargs) 2827 with self._lock: ~/Desktop/Spoken_Keyword_Spotting/newenv/lib/python3.6/site-packages/tensorflow/python/eager/function.py > 807 return self._stateless_fn(*args, **kwds) # pylint: disable=not-callableĨ09 # Release the lock early so that multiple threads can perform the call > 780 result = self._call(*args, **kwds)ħ82 new_tracing_count = self._get_tracing_count()Ĩ05 # In this case we have created variables on the first call, so we run theĨ06 # defunned version which is guaranteed to never create variables. ~/Desktop/Spoken_Keyword_Spotting/newenv/lib/python3.6/site-packages/tensorflow/python/eager/def_function.py > 1098 tmp_logs = train_function(iterator) 1099 if data_handler.should_sync: 1100 Use_multiprocessing) 1096 batch_size=batch_size):ġ097 callbacks.on_train_batch_begin(step) Validation_batch_size, validation_freq, max_queue_size, workers, Sample_weight, initial_epoch, steps_per_epoch, validation_steps, Validation_split, validation_data, shuffle, class_weight, In fit(self, x, y, batch_size, epochs, verbose, callbacks, > 108 return method(self, *args, **kwargs)ġ10 # Running inside run_distribute_coordinator already. In _method_wrapper(self, *args, **kwargs)ġ06 def _method_wrapper(self, *args, **kwargs):ġ07 if not self._in_multi_worker_mode(): # pylint: disable=protected-access ~/Desktop/Spoken_Keyword_Spotting/newenv/lib/python3.6/site-packages/tensorflow/python/keras/engine/training.py > 1 history=model.fit(x_tr, y_tr ,epochs=100, callbacks=, batch_size=64, validation_data=(x_val,y_val)) InvalidArgumentError Traceback (most recent call Mc = ModelCheckpoint('best_model.hdf5', monitor='val_acc', verbose=1, save_best_only=True, mode='max') Model = modelcreator.getmodel(modelname, input_shape, numclass)Įs = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=10, min_delta=0.0001) Here are my train and validation datasets dimensions: print(x_tr.shape) -> (17049, 39, 101, 1) Would be great if anyone could help with any suggestions to get it fixed. I believe this is happening before the validation step starts. My callbacks are quite simple ones related to model checkpoint and earlystopping as displayed below.My DB is from Google Speech dataset v0.01. I don't know why this happening in the last step only. In the very last step of the first epoch, I am facing an error as listed below, that suggests my tensor shape is changing somehow. After using the feature extractor and getting dimensions of the train and validation datasets with the compatible dimensions- (total_samples,39,101,1), I tried to train the model using model.fit.
I was trying to convert a custom DNN network originally written in Pytorch to Tensorflow 2 that has input shape required as (batchsize,39,101,1).
0 kommentar(er)
