KempnerForge
PyTorch-native framework for fault-tolerant distributed training of foundation models on AI clusters.
Repository: github.com/KempnerInstitute/KempnerForge
Portfolio
KempnerPulse
Real-time terminal dashboard for NVIDIA DCGM GPU metrics on HPC clusters, SLURM-aware and installable from PyPI.
Repository: github.com/KempnerInstitute/kempnerpulse
MLPerf Inference v6.1 submission
Llama 3.1-8B in FP8 on a single NVIDIA H200 with TensorRT-LLM, Closed division, Datacenter category.
Repository: mlcommons/inference_results_v6.1, closed/Naeem Khoshnevis