Hello,

I have troubles when running your program.

I have installed tensorflow 1.2 as required, but also with newer tensor flow versions it fails.
The module has only the interface "conv1d", which expects some other parameters, so this also does not lead to a solution.

Do you have any idea?

Thanks in advance

Traceback (most recent call last):
  File "main.py", line 103, in <module>
    main(args, prt)
  File "main.py", line 58, in main
    is_train=args['is_train'])
  File "/Projects/vrp/model/attention_agent.py", line 42, in __init__
    _scope=_scope+'Actor/')
  File "/Projects/vrp/shared/embeddings.py", line 28, in __init__
    self.project_emb = tf.layers.Conv1D(embedding_dim,1,
AttributeError: module 'tensorflow.python.layers.layers' has no attribute 'Conv1D'

I have run your program for around 82k epochs for VRP10. After each 200 epoch your program validated by 1000 test samples and according to the result ,which is average tour length of those 1000 test data, it is very clear that over the epochs result is improving. Its start with a value of 8.309866905212402 and the final validation result is 4.853826522827148 (in greedy method). But the problem is that after 82k epochs I have load the saved weight and only perform test on those 1000 data it produce a result with the value 7.8747758865356445. I have performed the test operation for several time but each and every time I got the same kind of result.
I am attaching the text files for your reference. Waiting for your answer.
Thank you
results.txt
results.txt

Hello,

First, thank you for the code, it's very interesting and usefull, i've also read your paper and i'm interested in TSP and VRP problem.

I want to use this model but with a "real" dataset. The point is that if I want to respect all distances between my "n" cities, I have to project them into dimensions "n-1".

I've seen the next parameter in your Env class (both TSP and VRP) :

• args['input_dim']: dimension of the problem which is 2

I want to change it (it's set into task_specific_params.py) to solve "real world" problem.

Have you tried change this into "n-1" dimension (9 for TSP 10) to maintain all distances you have between each pair of the 10 nodes ?

I'm triing to use your code, to adapt it to "real" problems, in which you can not consider distances as the crow flies. The real agent need to follow roads.

I'm triing to work on such an issue, by generating new structure of datas to train and test in some new dimensions.

Maybe you also did ?

I've work on a paper and I know that it's dificult to adapt some code made to illustrate progress in a field to an industrial real world big problem.

I'll appreciate if you could tell me if you have been considering my type of approach.

Thanks again !

Hello,

problem is about the valid data set. I found that some instance in the valid set is similar to some instance in the training set. Because there coordination are the same, and only the demand is different. Do you think it will have influence on the reliability of the result?

thank you very much!!!

Hi, thanks for your work!
I wonder if stochastic VRP mentioned in your paper is implemented in this project. I'm working on pickup-and-delivery with dynamic orders！
Thanks！

I am trying to run this on windows machine with just CPUs available, but it gives error
ModuleNotFoundError: No module named 'numpy.core._multiarray_umath'
ImportError: numpy.core.multiarray failed to import

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "", line 980, in _find_and_load
SystemError: <class '_frozen_importlib._ModuleLockManager'> returned a result with an error set
ImportError: numpy.core._multiarray_umath failed to import
ImportError: numpy.core.umath failed to import
2019-06-22 03:38:35.403667: F tensorflow/python/lib/core/bfloat16.cc:675] Check failed: PyBfloat16_Type.tp_base != nullptr

When I manually import numpy.core.multiarray and numpy.core.umath from python prompt it is importing fine.

Am I missing something?

Hello!
I appreciate your excellent work very much, so I tried to run your code in order to find about your result precisely.But as I trained,in the size 20,50,100，after a few steps,actor loss became NAN and critic loss became 0. Why does that happened?

Are the glimpses in the code corresponding to the equation 4 and 5 in the article?
I an quite confused of it.
if so, the M in the equation 5 is the number of glimpses?
I like your paper very much, so I want to know how to realize it.
Thank you ~