"mm = TrajPatternMiner(traj_list, learning_rate, training_epochs, \ display_step, units, batch_size, max_n_steps, frame_dim, window_size)"

[whalienvita]

When i run the code "mm = TrajPatternMiner(traj_list, learning_rate, training_epochs,
display_step, units, batch_size, max_n_steps, frame_dim, window_size)"

There's an error:
TypeCheckError Traceback (most recent call last)
Cell In[10], line 1
----> 1 mm = TrajPatternMiner(traj_list, learning_rate, training_epochs,
2 display_step, units, batch_size, max_n_steps, frame_dim, window_size)

File D:\Users\duhit\Desktop\323\FYP_trajectory_clustering-main\FYP_trajectory_clustering-main\model_2\traj_pattern_miner.py:20, in TrajPatternMiner.init(self, traj, learning_rate, training_epochs, display_step, units, batch_size, max_n_steps, frame_dim, window_size)
18 self.window_size = window_size
19 self.txc = TrajectoryBehaviorExtraction()
---> 20 self.t2v = Traj2Vec(self.learning_rate, self.training_epochs, self.display_step, self.units, self.batch_size, self.frame_dim)

File D:\Users\duhit\Desktop\323\FYP_trajectory_clustering-main\FYP_trajectory_clustering-main\model_2\traj2vec.py:27, in Traj2Vec.init(self, learning_rate, training_epochs, display_step, units, batch_size, frame_dim)
25 self.frame_dim = frame_dim
26 self.encoder = Encoder(self.units, self.batch_size)
---> 27 self.decoder = Decoder(self.units, self.batch_size, self.max_n_steps)
28 self.optimizer = tf.keras.optimizers.experimental.RMSprop(self.learning_rate)
29 self.checkpoint_dir = './training_checkpoints'

File D:\Users\duhit\Desktop\323\FYP_trajectory_clustering-main\FYP_trajectory_clustering-main\model_2\decoder.py:21, in Decoder.init(self, dec_units, batch_sz, max_length_input, attention_type)
18 self.sampler = tfa.seq2seq.sampler.TrainingSampler()
20 # Create attention mechanism with memory = None
---> 21 self.attention_mechanism = self.build_attention_mechanism(self.dec_units,
22 None, self.batch_sz*[self.max_length_input], self.attention_type)
24 # Wrap attention mechanism with the fundamental rnn cell of decoder
25 self.rnn_cell = self.build_rnn_cell(batch_sz)

File D:\Users\duhit\Desktop\323\FYP_trajectory_clustering-main\FYP_trajectory_clustering-main\model_2\decoder.py:46, in Decoder.build_attention_mechanism(self, dec_units, memory, memory_sequence_length, attention_type)
44 return tfa.seq2seq.BahdanauAttention(units=dec_units, memory=memory, memory_sequence_length=memory_sequence_length)
45 else:
---> 46 return tfa.seq2seq.LuongAttention(units=dec_units, memory=memory, memory_sequence_length=memory_sequence_length)

File D:\duhit\anaconda\envs\FYP\lib\site-packages\tensorflow_addons\seq2seq\attention_wrapper.py:543, in init(self, units, memory, memory_sequence_length, scale, probability_fn, dtype, name, **kwargs)
540 # For LuongAttention, we only transform the memory layer; thus
541 # num_units must match expected the query depth.
542 self.probability_fn_name = probability_fn
--> 543 probability_fn = self._process_probability_fn(self.probability_fn_name)
545 def wrapped_probability_fn(score, _):
546 return probability_fn(score)

File D:\duhit\anaconda\envs\FYP\lib\site-packages\typeguard_functions.py:264, in check_variable_assignment(value, expected_annotations, memo)
262 iterated_values.append(obj)
263 try:
--> 264 check_type_internal(obj, expected_type, memo)
265 except TypeCheckError as exc:
266 exc.append_path_element(argname)

File D:\duhit\anaconda\envs\FYP\lib\site-packages\typeguard_checkers.py:671, in check_type_internal(value, annotation, memo)
668 return
670 if not isinstance(value, origin_type):
--> 671 raise TypeCheckError(f"is not an instance of {qualified_name(origin_type)}")

TypeCheckError: probability_fn is not an instance of str

Who knows the solving methods? Emergency!

[AgnesYe1029]

When I was running on Local Jupyter notebook, there is no such error message. Are you running on Google Colab?

[whalienvita]

i am also on the Local Jupyter notebook, however there's an error. i straightly run the original code, do i need to edit it?

[AgnesYe1029]

Here we are not even touching probability_fn argument in our code though. Based on the trace you pasted, it seems a type check dependency issue inside site-packages\tensorflow_addons\seq2seq\attention_wrapper.py.

Updates: Kindly checkout this issue raised in tfa's source code. I think it's the same as your situation. While that issue still remains open at tfa side, here's a solution to bypass the type check: "comment out @TypeChecked inside LuongAttention(AttentionMechanism). "

[whalienvita]

Thank you very much! I removed the code containing TypeCheckError from site-packages\typeguard_checkers.py and the problem was solved.

[AgnesYe1029]

While removing TypeCheckError from site-packages\typeguard_checkers.py works, it might not be a sustainable practice as TypeCheckError supports type consistency for all .py documents.
Removing @TypeChecked annotation within site-packages\tensorflow_addons\seq2seq\attention_wrapper.py would be a better option with less impact.