OMA Skills Hub by OMA · Open Modus Accelerator
🎯 Use Case Hermes Skill

distributed-llm-pretraining-torchtitan

Pretrain LLMs at scale with PyTorch 4D parallelism.

仓库: NousResearch/hermes-agent · v1.0.1 · by Orchestra Research · MIT

Model ArchitectureDistributed TrainingTorchTitanFSDP2Tensor ParallelPipeline ParallelContext ParallelFloat8LlamaPretraining
文件路径
optional-skills/mlops/torchtitan/SKILL.md

查看源文件 (raw) ↗