1 indexed skills ยท max 10 per page
davila7/claude-code-templates ยท AI/ML
Megatron-Core trains LLMs from 2B to 462B parameters with up to 47% Model FLOP Utilization on H100 GPUs through advanced parallelism strategies.