Add MLPerf Llama 3.1 8B Recipes

[nv-mollys]

What does this PR do ?

Cherry-pick the MLPerf Llama 3.1 8B GB200/GB300 performance recipes from #5861 onto main and pin their layouts with unit tests.

Changelog

• Add GB200 Llama 3.1 8B perf recipes: llama31_8b_pretrain_8gpu_gb200_nvfp4_config, llama31_8b_pretrain_72gpu_gb200_nvfp4_config, llama31_8b_pretrain_512gpu_gb200_fp8cs_config.
• Add GB300 Llama 3.1 8B perf recipes: llama31_8b_pretrain_8gpu_gb300_nvfp4_config, llama31_8b_pretrain_72gpu_gb300_nvfp4_config, llama31_8b_pretrain_512gpu_gb300_fp8cs_config.
• Add _llama31_8b_common() in perf_recipes/llama/common.py for the shared MLPerf optimizer, DDP, model, scheduler, dataset, checkpoint, logger, and tokenizer defaults.
• Export the new recipe factories from perf_recipes/llama/__init__.py.
• Stop setting cfg.model.attention_backend = None on the library llama31_8b_pretrain_2gpu_h100_bf16_config.
• Add tests/unit_tests/recipes/test_llama31_8b_perf_recipes.py to pin parallelism, batch, LR, eval, CUDA graph, and precision for the six MLPerf recipes.
• Port-time fixes in the common helper: import torch for bfloat16 dtypes, and assign cfg.rng.data_parallel_random_init instead of annotating it.

Additional Information

• Related to Add MLPerf Llama 3.1 8B Recipes #5861

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[nv-mollys]

/ok to test 805d564

[github-actions]

Automatic Claude reviews have been retired. To request a pull-request review, post a comment containing:

/review

Add model=claude to use a Claude reviewer (the default is model=codex). mode=light|strict selects the review depth; for example, /review model=claude mode=strict. Comment /review help for all options.