Popular repositories Loading

1. higgsfield higgsfield Public

   Orchestrate fault-tolerant, ultra-scalable GPU training for trillion-parameter LLMs with ZeRO-3 and PyTorch FSDP.

   Jupyter Notebook

2. pamene9.github.io pamene9.github.io Public

   Orchestrate fault-tolerant multi-node GPU training for trillion-parameter LLMs with elastic, queue-managed resource allocation.

   HTML