search

vita-epfl / CrowdNav

[ICRA19] Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learning
MIT License
560 stars 166 forks source link

Trainning problem, need help, SOS #69

Closed JiangxinkeTao closed 8 months ago

JiangxinkeTao commented 8 months ago

Dear Author, your paper is good, but i face some troubles, The following info is my problem, i hope that you can help me to solve the problem, thank you very much !!!

tomas@tomas:~/Documents/SocialNavi/CrowdNav/crowd_nav$ python3 train.py --policy sarl --gpu Output directory already exists! Overwrite the folder? (y/n)y 2023-10-25 23:09:22, INFO: Current git head hash code: %s 2023-10-25 23:09:22, INFO: Using device: cpu 2023-10-25 23:09:22, INFO: Policy: SARL w/ global state 2023-10-25 23:09:22, INFO: human number: 5 2023-10-25 23:09:22, INFO: Not randomize human's radius and preferred speed 2023-10-25 23:09:22, INFO: Training simulation: circle_crossing, test simulation: circle_crossing 2023-10-25 23:09:22, INFO: Square width: 10.0, circle width: 4.0 2023-10-25 23:09:22, INFO: Current learning rate: 0.010000 2023-10-25 23:13:50, INFO: TRAIN has success rate: 0.89, collision rate: 0.09, nav time: 12.18, total reward: 0.2416 Traceback (most recent call last): File "train.py", line 177, in main() File "train.py", line 130, in main trainer.optimize_epoch(il_epochs) File "/home/tomas/Documents/SocialNavi/CrowdNav/crowd_nav/utils/trainer.py", line 39, in optimize_epoch outputs = self.model(inputs) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl return self._call_impl(*args, kwargs) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl return forward_call(*args, *kwargs) File "/home/tomas/Documents/SocialNavi/CrowdNav/crowd_nav/policy/sarl.py", line 48, in forward scores = self.attention(attention_input).view(size[0], size[1], 1).squeeze(dim=2) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl return self._call_impl(args, kwargs) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl return forward_call(*args, kwargs) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/container.py", line 215, in forward input = module(input) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl return self._call_impl(*args, *kwargs) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl return forward_call(args, kwargs) File "/home/tomas/.local/lib/python3.8/site-packages/torch/nn/modules/linear.py", line 114, in forward return F.linear(input, self.weight, self.bias) RuntimeError: could not create a primitive descriptor for a matmul primitive

ChanganVR commented 8 months ago

I don't think is an issue with this library. Looks like some package (version) issues. Does pytorch even work in your environment? If yes, you can start by removing codes and pinpointing what exactly is the issue. Have you tried the online solutions? e.g., https://github.com/pytorch/pytorch/issues/89104

JiangxinkeTao commented 8 months ago

my previous python version is 3.8，i change it to 3.7，then the problem is solved, i forgot to close the issue , and thank you for your response sincerely.