# torchx **URL:** https://discuss.pytorch.org/c/torchx/35.md [Latest](https://discuss.pytorch.org/latest.md) · [Categories](https://discuss.pytorch.org/categories.md) --- ## [About the torchx category](https://discuss.pytorch.org/t/about-the-torchx-category/134813)
**Author:** [@dzhulgakov](https://discuss.pytorch.org/u/dzhulgakov)\ **Replies:** 0
TorchX is an SDK for quickly building and deploying ML applications from R&D to production. It offers various builtin components that encode MLOps best practices and make advanced features like distributed training and h… --- ## [Implementing NodeSelectors and Tolerations with TorchX on Kubernetes(EKS)](https://discuss.pytorch.org/t/implementing-nodeselectors-and-tolerations-with-torchx-on-kubernetes-eks/186437)
**Author:** [@vara\_bonthu](https://discuss.pytorch.org/u/vara_bonthu)\ **Replies:** 3\ **Last updated:** [May 9, 2025, 5:17pm UTC](https://discuss.pytorch.org/t/implementing-nodeselectors-and-tolerations-with-torchx-on-kubernetes-eks/186437 "2025-05-09T17:17:30Z")
I’m currently working with TorchX in conjunction with Volcano scheduling for my training jobs on an Amazon EKS cluster. I’ve also integrated Karpenter autoscaler for effective node scaling. Additionally, I’m using manage… --- ## [Using any scheduler except local\_cwd evokes FileNotFoundError](https://discuss.pytorch.org/t/using-any-scheduler-except-local-cwd-evokes-filenotfounderror/216867)
**Author:** [@DougReeder](https://discuss.pytorch.org/u/DougReeder)\ **Replies:** 0\ **Last updated:** [February 19, 2025, 5:51am UTC](https://discuss.pytorch.org/t/using-any-scheduler-except-local-cwd-evokes-filenotfounderror/216867 "2025-02-19T05:51:37Z")
I’ve installed torchx in a venv using Python 3.12: (venv-vllm) ~/machine-learning/vllm $ python --version Python 3.12.9 (venv-vllm) ~/machine-learning/vllm $ pip install "torchx\[dev\]" ... Running a job locally works fi… --- ## [Torch 2 installed - Could not load library libcudnn\_cnn\_infer.so.8. Error: libnvrtc.so: cannot open shared object file: No such file or directory](https://discuss.pytorch.org/t/torch-2-installed-could-not-load-library-libcudnn-cnn-infer-so-8-error-libnvrtc-so-cannot-open-shared-object-file-no-such-file-or-directory/176167)
**Author:** [@FurkanGozukara](https://discuss.pytorch.org/u/FurkanGozukara)\ **Replies:** 5\ **Last updated:** [April 1, 2023, 7:23pm UTC](https://discuss.pytorch.org/t/torch-2-installed-could-not-load-library-libcudnn-cnn-infer-so-8-error-libnvrtc-so-cannot-open-shared-object-file-no-such-file-or-directory/176167 "2023-04-01T19:23:08Z")
Hello. I have installed torch 2 via this command on runpod io instance everything installed perfectly fine pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 with torch 1 we w… --- ## [NVIDIA GeForce RTX 3090 with CUDA capability sm\_86 is not compatible with the current PyTorch installation](https://discuss.pytorch.org/t/nvidia-geforce-rtx-3090-with-cuda-capability-sm-86-is-not-compatible-with-the-current-pytorch-installation/141940)
**Author:** [@akslduk](https://discuss.pytorch.org/u/akslduk)\ **Replies:** 31\ **Last updated:** [December 8, 2022, 9:39am UTC](https://discuss.pytorch.org/t/nvidia-geforce-rtx-3090-with-cuda-capability-sm-86-is-not-compatible-with-the-current-pytorch-installation/141940 "2022-12-08T09:39:30Z")
Hello, I’m getting following error: NVIDIA GeForce RTX 3090 with CUDA capability sm\_86 is not compatible with the current PyTorch installation. The current PyTorch install supports CUDA capabilities sm\_37 sm\_50 sm\_60 s… --- ## [How to configure cmake while building pytorch from source code?](https://discuss.pytorch.org/t/how-to-configure-cmake-while-building-pytorch-from-source-code/166718)
**Author:** [@studyself](https://discuss.pytorch.org/u/studyself)\ **Replies:** 9\ **Last updated:** [November 24, 2022, 6:19am UTC](https://discuss.pytorch.org/t/how-to-configure-cmake-while-building-pytorch-from-source-code/166718 "2022-11-24T06:19:42Z")
HI, I need to use FBEGMM in pytorch, and I’m building pytorch from source code. The errors ask me to set cmake configuration as -DENABLE\_PRECOMPILED\_HEADERS=OFF. I modify cmake.py and try to add -DENABLE\_PRECOMPILED\_… --- ## [Stdlib.h not found while building pytorch from source code](https://discuss.pytorch.org/t/stdlib-h-not-found-while-building-pytorch-from-source-code/166667)
**Author:** [@studyself](https://discuss.pytorch.org/u/studyself)\ **Replies:** 0\ **Last updated:** [November 23, 2022, 1:49am UTC](https://discuss.pytorch.org/t/stdlib-h-not-found-while-building-pytorch-from-source-code/166667 "2022-11-23T01:49:04Z")
Hi, I got these errors while building pytorch from source code. /usr/include/c++/7/cstdlib:75:15: fatal error: stdlib.h: No such file or directory By following: I tried -DENABLE\_PRECOMPILED\_HEADERS=OFF CMAKE\_CFLAGS… --- ## [Looking for a practical example of making KFP app using torchx](https://discuss.pytorch.org/t/looking-for-a-practical-example-of-making-kfp-app-using-torchx/165008)
**Author:** [@helloybz](https://discuss.pytorch.org/u/helloybz)\ **Replies:** 0\ **Last updated:** [November 2, 2022, 8:57am UTC](https://discuss.pytorch.org/t/looking-for-a-practical-example-of-making-kfp-app-using-torchx/165008 "2022-11-02T08:57:26Z")
Hi. Where can I find a good practical example of building KubeFlow Pipeline app using torchx? I’ve been through the documents for torchx, and now I get how torchx work. However, I have no idea how to design my app wit… --- ## [Pytorch model training stops at the end of the first Epoch](https://discuss.pytorch.org/t/pytorch-model-training-stops-at-the-end-of-the-first-epoch/160618)
**Author:** [@rabrab](https://discuss.pytorch.org/u/rabrab)\ **Replies:** 1\ **Last updated:** [September 3, 2022, 7:56pm UTC](https://discuss.pytorch.org/t/pytorch-model-training-stops-at-the-end-of-the-first-epoch/160618 "2022-09-03T19:56:02Z")
Hello Everyone, I have seen related post about my issue, but after applying the advice and changing the num\_workers, my training still gets intrrupted without any error as shown in screenshot bellow, I have a Pytorch 1.… --- ## [How to add a new parameter which is derived from output of the previous layer of the neural network model](https://discuss.pytorch.org/t/how-to-add-a-new-parameter-which-is-derived-from-output-of-the-previous-layer-of-the-neural-network-model/158264)
**Author:** [@Hritik\_Gopal\_Shah](https://discuss.pytorch.org/u/Hritik_Gopal_Shah)\ **Replies:** 3\ **Last updated:** [August 8, 2022, 8:20pm UTC](https://discuss.pytorch.org/t/how-to-add-a-new-parameter-which-is-derived-from-output-of-the-previous-layer-of-the-neural-network-model/158264 "2022-08-08T20:20:41Z")
How to add a new parameter which is derived from output of the previous layer, apart from weights and baises in Custom layer of the neural network model. I want the ouput of previous layer rolled(torch.roll) by 1 positi… --- ## [ValueError: Only sparse matrices with 32-bit integer indices are accepted. Got int64 indices](https://discuss.pytorch.org/t/valueerror-only-sparse-matrices-with-32-bit-integer-indices-are-accepted-got-int64-indices/155944)
**Author:** [@cola888666](https://discuss.pytorch.org/u/cola888666)\ **Replies:** 2\ **Last updated:** [July 6, 2022, 8:05am UTC](https://discuss.pytorch.org/t/valueerror-only-sparse-matrices-with-32-bit-integer-indices-are-accepted-got-int64-indices/155944 "2022-07-06T08:05:07Z")
when training svm by sklearn with fashion\_mnist dataset . ValueError: Only sparse matrices with 32-bit integer indices are accepted. Got int64 indices. Can you help me? Thanks! --- ## [Error while saving state of optimizer in Pytorch](https://discuss.pytorch.org/t/error-while-saving-state-of-optimizer-in-pytorch/136463)
**Author:** [@Sumera\_Rounaq](https://discuss.pytorch.org/u/Sumera_Rounaq)\ **Replies:** 8\ **Last updated:** [November 17, 2021, 5:12am UTC](https://discuss.pytorch.org/t/error-while-saving-state-of-optimizer-in-pytorch/136463 "2021-11-17T05:12:03Z")
Hi all, I am saving state of optimizer and getting this error “state\_dict() missing 1 required positional argument: ‘self’”. Code: torch.save({ 'epoch': num\_epochs, 'model\_state\_dict': model.s…