A training framework for LLMs, VLMs, diffusion and embodied models on NVIDIA GPUs and Kunlun XPUs, built on Megatron-LM and torch-native backends.
69458PythonApache-2.0Updated 8 hours ago
#megatron
Open-source projects tagged “megatron”
Curated GitHub repositories related to megatron, with AI summaries and key stats.