List your company

LLM Training on Llama2-70B (64 x H100-8)

7.8Iteration Time (s)Megatron-LM
7.6888.4449.29.956Jul 7, 2025

Evaluation results

MethodLinks
Megatron-LMHardware=64 x H100-82025.077.847.2—
AXLearnHardware=64 x H100-82025.079.240—
MaxTextHardware=64 x H100-82025.079.439.1—
PyTorch FSDPHardware=64 x H100-82025.0710.634.7—