Update LLAMA3 70B LoRa Base Configs for GB300, GB200 and H100

[rhmukundan]

Summary by CodeRabbit

• Chores
  • Updated Llama 3 70B LoRA workload configurations for GB300, GB200, and H100 hardware platforms
  • Adjusted parallelism parameters (tensor, pipeline, and context) across model variants
  • Modified recomputation settings for H100 configurations

[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.

[coderabbitai]

📝 Walkthrough

Walkthrough

This change modifies Llama3 70B LoRA configuration settings for GB300, GB200, and H100 hardware platforms. Updates include adjusting tensor/pipeline/context parallelism dimensions and recomputation parameters across base configs and their variants through replace() calls.

Changes

Cohort / File(s)	Summary
Llama3 LoRA Performance Config Updates scripts/performance/configs/llama/llama3_workload_base_configs.py	Modified parallelism dimensions (tensor_model_parallel_size, pipeline_model_parallel_size, context_parallel_size, virtual_pipeline_model_parallel_size) for GB300, GB200, and H100 base configs. Updated public aliases to derive from refactored bases using replace() calls. Adjusted recompute_num_layers for H100_BF16_V1 from 2 to 1.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Suggested labels

Run CICD, performance

Suggested reviewers

• malay-nagda
• erhoo82
• thomasdhc

🚥 Pre-merge checks | ✅ 3 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Test Results For Major Changes	⚠️ Warning	PR contains major configuration changes for LLAMA3 70B LoRA model parallelism across multiple platforms but lacks comprehensive testing information, performance metrics, and validation details in the PR description.	Add test results validating configurations, before-and-after performance metrics for each platform, batch sizes used, and specify which configurations have been validated.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately and specifically describes the main change: updating LLAMA3 70B LoRa base configurations for three hardware platforms (GB300, GB200, H100).
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch rmukundan/llama3_lora_baseconfigs_change

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[rhmukundan]

/ok to test 16aa00e