Hi , I'm vishnu
Can anyone please share or tell me how to train pytorch CNN models with FPGA or GPU or VPU's on Intel DevCloud's Jupyter lab!!!?
Any basic example with code is enough.
Thanks for reaching out.
Could you please confirm which DevCloud are you using? Is it FPGA DevCloud (FPGA acceleration) / oneAPI devcloud (datacentre workloads)?
If you are using FPGA devcloud and has queries regarding FPGA, please post your query in https://forums.intel.com/s/topic/0TO0P0000001AUUWA2/intel-high-level-design?language=en_US Forum.
If you would like to train your workload using CPU ,you can choose oneAPI devcloud.The oneAPI Devcloud has preinstalled oneAPI AI Analytics toolkit which contains popular DL frameworks like tensorflow and pytorch.
To utilize the preinstalled AI Analytics toolkit in devcloud follow the steps mentioned in the below documentation
You can find a code sample in this link https://github.com/intel/AiKit-code-samples/tree/master/PyTorch_HelloWorld
Thank you for the above links, Can you please tell me how to train my workload in pytorch on GPU's, VPU's on oneAPI Devcloud
Thanks for the update.
The oneAPI devcloud (datecentre cloud) has only CPU's and iGPU. You can try pytorch training in CPU in oneAPIDevcloud but the pytorch support for igpu is currently not available.
For simple training workload you can try our hello world example shared earlier. You can explore more samples in the pytorch official site https://pytorch.org/tutorials/index.html#
If you want to try inference ,then our devcloud for edge has all the options to test the performance on CPU, GPU, VPU, & FPGA Architectures.You can post any queries regarding that in our devcloud for edge forum (https://software.intel.com/en-us/forums/intel-devcloud-for-edge)
Hope we have answered your questions.Could you please give us an update?