Deleted user
Deleted user
@firesurfer Thanks. Then I will need to rely on ros2control more.
Hey @wopian! One of my PR's is active but I'd like to start work on this if that's okay. Can I be assigned? :)
@wopian Created two PRs for this issue as specified in the description.
Thanks to the various CDVD timing changes, this is not an issue anymore, the only remaining gremlin is this log: 
See #684.
Me too.
import torch_xla torch_xla.__version__ 2.4.0+libtpu
@zpcore than you, I see the code running on all tpu core on kaggle v3-8 now. But the new problem is after torch_xla.distributed.xla_multiprocessing.spawn(process, start_method='fork') I just can see one process...
@zpcore I figure out the problem, the StableDiffusionControlNetPipeline is very big and occupy many memory. If I use 8 process, I create 8 StableDiffusionControlNetPipeline pipeline. So that they eat up...
@zpcore the following code ``` import torch_xla, diffusers, builtins, imageio, os, PIL.Image, controlnet_aux, sys, torch os.environ.pop('TPU_PROCESS_ADDRESSES') reader = imageio.get_reader('/kaggle/input/controlnet/pose.mp4', 'ffmpeg') openpose = controlnet_aux.DWposeDetector(det_config='yolox_l_8xb8-300e_coco.py', pose_config='dwpose-l_384x288.py') poses = [openpose(PIL.Image.fromarray(reader.get_data(_)).resize((512, 768))) for _...