mlcommons
SDXL model download link broken
https://cloud.mlcommons.org/index.php/s/LCdW5RM6wgGWbxC https://cloud.mlcommons.org/index.php/s/DjnCSGyNBkWA4Ro Pages Not Found error for SDXL model links.
Not only SDXL but all MLCommons hosted links like GPTJ, DLRMv2 are also broken via wget. MLCommons admins have suggested to use the rclone link which are wrapped in the CM comment here
cm pull repo ctuning@mlcommons-ck
Does every submitter need to use the ctuning account or they can use their own accounts?
@nvyihengz Please use cm pull repo mlcommons@ck - the ctuning one was a temporal link until the PR was merged.
mlcommons@ck is a short hand for https://github.com/mlcommons/ck
Known affected workloads @mrmhodak @pgmpablo157321 SDXL https://github.com/mlcommons/inference/tree/master/text_to_image GPTJ https://github.com/mlcommons/inference/tree/master/language/gpt-j DLRMv2 https://github.com/mlcommons/inference/tree/master/recommendation/dlrm_v2/pytorch#downloading-model-weights
@nvyihengz We do have the cloudflare rclone commands for all. (wget links are still broken though)
python3 -m pip install cmind
cm pull repo mlcommons@ck
cm run script --tags=get,ml-model,gptj,_pytorch,_rclone -j
cm run script --tags=get,ml-model,dlrm,_pytorch,_weight_sharded,_rclone -j
cm run script --tags=get,ml-model,sdxl,_fp16,_rclone -j
cm run script --tags=get,ml-model,sdxl,_fp32,_rclone -j
Running cm command will print the underlying rclone config but it is better to use cm commands as the configs can change later but cm commands won't.