Pending tasks for Inference v4.1

[anandhu-eng]

---

Migrated from mlcommons/mlperf-automations_archived#20
Originally created by @arjunsuresh on Thu, 18 Jul 2024 08:35:39 GMT

---

• Criteo preprocessed dataset is now shared by MLC. So, we need to update this script to automate the DLRM data creation for Nvidia MLPerf inference implementation cm4mlops#179
• Fix broken URLs #26
• Suggestions needed - better handling of backup URLs #25
• Create a way to make CM Nvidia docker container handle dependencies of all the benchmarks so that the docker remains the same for all the benchmarks cm4mlops#174
• Do a test submission to github for Intel and Nvidia implementations cm4mlops#170
• Update the automatically generated submission README to include model precision info cm4mlops#173
• Create a separate CM script to get MLC rclone config as this config is used by multiple CM scripts cm4mlops#172
• Check if a user has sudo permission and if not skip the commands requiring sudo with a warning cm4mlops#168
• Support for cleaning downloaded models and preprocessed datasets for Nvidia implementation #24
• Add dependency diagrams to CM script readmes cm4mlops#171
• Add dependency diagrams to MLPerf inference submission readme's cm4mlops#178
• Add checksum checks for all MLPerf inference dataset and model downloads cm4mlops#167
• Cuda installation inside CM cache is problematic especially in multi user setup #23
• Add documentation page for CM MLPerf inference v4.1 reproducibility #22
• Add a visualizer running on Github for MLPerf inference results cm4mlops#206