Core workflows and day-to-day usage.
- Primus tools: start here—an at-a-glance catalog of all Primus tools and ecosystem projects with how-to starting points
- CLI reference:
primus-climodes, flags, and subcommands - Configuration system: YAML configuration model, presets, overrides, inheritance
- Environment and XLA flags: where to set env vars, how to override
XLA_FLAGSsafely, how to verify what took effect - Pretraining: pretraining concepts: backends, YAML structure, parallelism, configuration inventory
- End-to-end training recipes: pretraining commands: copy-paste, GPU-arch-specific run commands
- Megatron-LM training performance validation: reproduce the published Megatron backend benchmarks on the
rocm/primusimage - TorchTitan training performance validation: reproduce the published TorchTitan backend benchmarks on the
rocm/primusimage - JAX MaxText training performance validation: reproduce the AMD-published MaxText benchmarks via Primus, MAD, or the standalone scripts
- Post-training: SFT and LoRA fine-tuning via Megatron Bridge
- Node-smoke test instruction: screen a cluster fast and exclude bad nodes before launching a real training job
- Preflight: cluster diagnostics and environment validation
- Run preflight without a container: run cluster-diagnostic tool directly on the host
- Micro-benchmarking: GEMM, RCCL, and dense-GEMM benchmark suites
- Projection: memory and performance projection tools
- Tuning agent: LLM-driven search for an optimal training configuration (uses projection as an oracle)