search

Lifelong-Robot-Learning / LIBERO

Benchmarking Knowledge Transfer in Lifelong Robot Learning
MIT License
171 stars 29 forks source link

Calling env.reset() multiple times causing kernel crash #24

Closed JummerCloth closed 1 week ago

JummerCloth commented 3 weeks ago

Hello authors! Thanks for this wonderful contribution to the robot learning community. I have been trying to use this simulator for some evaluation on a model I trained, but the system seems to bug out every time after a while. Later investigation shows that after calling env.reset() around 20 times, it will never succeed, resulting in infinite ValueError. I have reproduced a similar issue with a small modification of the getting started code provided on your README file: dummy.py.zip

With this simple script, the terminal keeps outputting env reset after around 20 iterations. Could you please look into this issue? Much thanks! I am using the libero env provided, with the following machine specifications: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian Address sizes: 43 bits physical, 48 bits virtual CPU(s): 96 On-line CPU(s) list: 0-95 Thread(s) per core: 2 Core(s) per socket: 24 Socket(s): 2 NUMA node(s): 2 Vendor ID: AuthenticAMD CPU family: 23 Model: 49 Model name: AMD EPYC 7402 24-Core Processor Stepping: 0 Frequency boost: enabled CPU MHz: 1497.918 CPU max MHz: 2800.0000 CPU min MHz: 1500.0000 BogoMIPS: 5599.99 Virtualization: AMD-V L1d cache: 1.5 MiB L1i cache: 1.5 MiB L2 cache: 24 MiB L3 cache: 256 MiB NUMA node0 CPU(s): 0-23,48-71 NUMA node1 CPU(s): 24-47,72-95

JummerCloth commented 1 week ago

I resolved the issue by upgrading the robosuite: pip install --force-reinstall robosuite, and downgrading numpy to a 1.x version.

pengzhi1998 commented 1 day ago

Hi, previously I didn't have this issue. But this week it suddenly happened to me as well which is very wired. Thank you for your solution! However, may I ask which Numpy version you are using specifically?