graph_comb_opt errors on training

when I run ./run_nstep_dqn.sh in s2v_tsp2d it returns

and some problems also happen when I run ./run_nstep_dqn in s2v_mvc

can you help me ??

Jun 07 '19 08:06 YujiaoHu

Hi there,

Have you built the .so file successfully?

Jun 07 '19 20:06 Hanjun-Dai

I tried again, and I built the .so successfully. I am sorry I forget to build it in the tsp2d. But the second problem still exists.
File "/MTSP/DQN/code/s2v_tsp2d/tsp2d_lib/tsp2d_lib.py", line 17, in init arr[:] = args TypeError: bytes or integer address expected instead of str instance

I run it in the nvidia-docker, and use cuda8.0 and nvcc8.0

Jun 08 '19 04:06 YujiaoHu

which version of python are you using? I'm using py2 in this repo

Jun 08 '19 16:06 Hanjun-Dai

I used python3 previously, and now I change to py2. The error is solved. Thank you. I can run your code, now.

Jun 09 '19 09:06 YujiaoHu

hi,i also meet the problem ,i want to know how to built .so file

Nov 20 '19 14:11 no-wings

go to clib folder under each experiment, and copy the Makefile.example to Makefile (with necessary adaptation to your system config). After that simply run make command.

Nov 29 '19 22:11 Hanjun-Dai