NVIDIA-NeMo / NVIDIA-NeMo/RL

Synthetic rollout length for GRPO performance benchmarking

Open
#1,302 0 comments 1 reaction 1 assignee Claimed by @guyueh1 View on GitHub
enhancement Performance
Dominant language
Python
Stars
2k
Forks
561
Avg merge
4d 5h
Merged PRs (30d)
145

Description

**Is your feature request related to a problem? Please describe.**
Vllm benchmarking CLI supports ignoring EOS and synthesizing given OSL ([--random-output-len ](https://docs.vllm.ai/en/latest/cli/bench/serve.html?h=ignore_eos#-random-output-len)). Can we use this feature to generate synthetic benchmarks of fixed OSL for evaluating perf at certain sequence length scenario?

**Describe the solution you'd like**
A clear and concise description of what you want to happen.

**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.

**Additional context**
Add any other context or screenshots about the feature request here.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.