yunzhuli / dpi-net Goto Github PK

I try to install repo and clone PyFlex submodule like in instruction, but always get mistake

Submodule 'PyFleX' ([email protected]:YunzhuLi/PyFleX.git) registered for path 'PyFleX'
Cloning into '/content/DPI-Net/PyFleX'...
Host key verification failed.
fatal: Could not read from remote repository.

Please make sure you have the correct access rights
and the repository exists.
fatal: clone of '[email protected]:YunzhuLi/PyFleX.git' into submodule path '/content/DPI-Net/PyFleX' failed
Failed to clone 'PyFleX'. Retry scheduled
Cloning into '/content/DPI-Net/PyFleX'...
Host key verification failed.
fatal: Could not read from remote repository.

Please make sure you have the correct access rights
and the repository exists.
fatal: clone of '[email protected]:YunzhuLi/PyFleX.git' into submodule path '/content/DPI-Net/PyFleX' failed
Failed to clone 'PyFleX' a second time, aborting

Hi, I installed DPI-Net and dependencies as outlined in README.md file. As I have cuda 10.2, I compiled PyFlex in docker (as suggested). But when I try to run a sample, I get the following error. Any idea/suggestions how I can resolve this?

$ python test_FluidFall.py
Unable to initialize SDLCould not initialize GL extensions
*** buffer overflow detected ***: python terminated
Aborted (core dumped)

$ bash scripts/eval_FluidFall.sh
Namespace(attr_dim=1, dataf='data/data_FluidFall', debug=0, dt=0.016666666666666666, env='FluidFall', epoch=12, eval=1, evalf='dump_FluidFall/eval_FluidFall', iter=180000, n_his=0, n_instance=1, n_instances=0, n_stages=1, neighbor_radius=0.08, nf_effect=200, nf_particle=200, nf_relation=300, outf='dump_FluidFall/files_FluidFall', position_dim=3, pstep=2, relation_dim=1, shape_state_dim=14, state_dim=6, time_step=121, time_step_clip=5, verbose_data=0, verbose_model=0)
Loading stored stat from data/data_FluidFall
Loading network from dump_FluidFall/files_FluidFall/net_epoch_12_iter_180000.pth
Unable to initialize SDLCould not initialize GL extensions
*** buffer overflow detected ***: python terminated
scripts/eval_FluidFall.sh: line 1: 18298 Aborted                 (core dumped) python eval.py --env FluidFall --epoch 12 --iter 180000

Traceback (most recent call last):
File "test_FluidFall.py", line 3, in
import pyflex
ImportError: No module named pyflex

But pyflex should have been installed?Can you help me deal with it ？

Hi,

I am wondering if you could help me with this - We got a segfault when running python test_FluidFall.py:

Unable to initialize SDLCould not initialize GL extensions
Segmentation fault (core dumped)  

Environment settings:

• docker container is based on: nvidia/cuda:9.1-devel-ubuntu16.04.
• Python 3.7
• Pytorch 1.0.0

Thanks!
Xin

I get the following PyTorch Gradient Error:

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [200, 6]], which is output 0 of AsStridedBackward0, is at version 2; expected version 1 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later. Good luck!

which I narrowed down to the following line in models.py. By changing it from return self.linear_2(x) to return self.linear_2(x.clone().detach()) in the ParticlePredictor class, I get rid of this gradient error, but now I cannot reproduce the results from the paper for RiceGrip. I'm not sure if detaching this variable from the gradient computation graph is causing this issue, but I'm not sure how else to get this code to run. Are there any solutions? FYI I'm running PyTorch 2.2.1

Dear Yunzhi Li,

I appreciate your generosity in sharing the source code.

I am currently endeavoring to replicate the original datasets on a remote headless server running Linux 18.04 with Nvidia Driver, and this setup lacks physical monitors. It appears that OpenGL doesn't work in the absence of a physical screen.

Could you kindly provide more information about your server environment? Specifically, does it include a physical monitor? I am eager to determine whether it is feasible to reproduce the dataset within a screenless environment.

Best,
Tommy

Hi, installation guide for PyFlex explains installing on a GPU machine. Is it possible to install and run this (evaluation only) on a CPU?