Shriya Rishab

icon company Nvidia icon location Senior Deep Learning Engineer @ Nvidia. MS in CS @ Columbia University

Results 24 comments of Shriya Rishab

The default training script of DLRM v2 does not reach the reported AUC.

@Kevin0624 has this been resolved? Closing as it has been more than a year since the last activity

Bert pretrain script message "Could not find trained model in model_dir: /tmp/output/"

@mahmoodn is this still an issue?

Is the Image Classification benchmark ResNet-50 v1 or ResNet-50 v1.5?

Closing as the benchmark is dropped from Training benchmark suite

added train_samples keyword for compliance check

@itayhubara @sgpyc - please review

does not have storage.objects.list access to the Google Cloud Storage bucket

All required data can be downloaded using instructions in the [S3 artifacts download section](https://github.com/mlcommons/training/blob/master/large_language_model/megatron-lm/README.md#s3-artifacts-download) of the README.

Checkpointing DLRMv2

@mailvijayasingh is this still an issue?

Checkpointing DLRMv2

Closing as it has been more than a year since the last comment

Are gpt tokenizer model open-source?

All required data can be downloaded using instructions in the [S3 artifacts download section](https://github.com/mlcommons/training/blob/master/large_language_model/megatron-lm/README.md#s3-artifacts-download) of the README.

RNNT apt-get update Fails in Docker Build

Closing based on [comment](https://github.com/mlcommons/training/issues/585#issuecomment-1381043243)

Unable to run unit tests of distributed checkpointing in Megatron-LM

@MingjiHan99 can you share more details and reproduction steps if this is still an issue?