- 35157 Views
- 2 replies
- 1 kudos
Resolved! torch.cuda.OutOfMemoryError: CUDA out of memory
Hi,I am using pynote/whisper large model and trying to process data using spark UDF and getting following error.torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 172.00 MiB (GPU 0; 14.76 GiB total capacity; 6.07 GiB already allocated...
- 35157 Views
- 2 replies
- 1 kudos
- 1 kudos
Try to run these codesimport torchtorch.cuda.empty_cache()And make sure to find the optimize batch size otherwise the error can occur again
- 1 kudos
- 1375 Views
- 1 replies
- 0 kudos
Does Databricks Container Services (DCS) support for GPU containers with Databricks Runtime 11.3 LTS and higher?
I have been trying to start a cluster using DCS with GPU containers (https://github.com/databricks/containers/tree/master/ubuntu/gpu), but was only successful with Databricks Runtime 10.4 LTS and lower. With Databricks Runtime 11.3 LTS and higher, I ...
- 1375 Views
- 1 replies
- 0 kudos
- 0 kudos
Hello @ppang ! Since you posted your question, the repository you shared has received an update, which includes the following warning: "Using conda in DCS images is no longer supported starting Databricks Runtime 9.0. We highly recommend users to ext...
- 0 kudos
- 2030 Views
- 2 replies
- 0 kudos
Running Keras model training with HorovodRunner works until the training function is exited ("The MPI_Query_thread() function was called after MPI_FINALIZE was invoked.")
I am running training of a Keras/Tensorflow deep learning model on a cluster of (for now) 2 workers and 1 driver (T4 GPU, 28GB, 4 core) using the Databricks provided HorovodRunner. It all seems to go well and the performance scales quite nicely over ...
- 2030 Views
- 2 replies
- 0 kudos
- 0 kudos
I personally suspect it's your callbacks. Can you remove all those state callbacks and see if that is it?
- 0 kudos
- 3270 Views
- 2 replies
- 1 kudos
Model serving with GPU cluster
Hello Databricks community!We are facing a strong need of serving some of public and our private models on GPU clusters and we have several requirements:1) We'd like to be able to start/stop the endpoints (best with scheduling) to avoid excess consum...
- 3270 Views
- 2 replies
- 1 kudos
- 1 kudos
Hi @Alisher Akh​ Does @Debayan Mukherjee​'s answer help? If yes, would you be happy to mark the answer as best so that other members can find the solution more quickly? If not, please tell us so we can help you further. Cheers!
- 1 kudos
- 1858 Views
- 2 replies
- 2 kudos
Why is GPU accelerated node much slower than CPU node for training a random forest model on databricks?
I have a dataset about 5 million rows with 14 features and a binary target. I decided to train a pyspark random forest classifier on Databricks. The CPU cluster I created contains 2 c4.8xlarge workers (60GB, 36core) and 1 r4.xlarge (31GB, 4core) driv...
- 1858 Views
- 2 replies
- 2 kudos
