Software Development
Inspect Hugging Face decoder layers on meta tensors and plan or run per-rank BF16, FP8, and NVFP4 GEMM or fused-MoE microbenchmarks with the bundled scripts and a local FlashInfer checkout. Use when choosing a model, GPU, TP, EP, or M/token-concurrency sweep; deriving common fused QKV and gate/up shapes without loading checkpoint weights; or using FlashInfer benchmark utilities. Do not use for end-to-end server throughput or request latency.
Publisher / source
NVIDIA/Model-Optimizer (found on GitHub top repositories)
Upstream record
Version
Version 1
Published
2026-11-13
Publication review
Approved for publication
Content fingerprint
8ab731076bccb431776cc6255b909b4da3561e317803f9364f4c0b3d9fed60f9
Review before you use it
Publication approval is a recorded review state, not a guarantee of safe execution. Check the upstream license, required tools and permissions, and test in your intended client.
Sign in to inspect the full instructions and download the SKILL.md file. This public page contains metadata only.
Source: https://prompeteer.ai/skills-hub/890e77f7-d57b-4312-8cf7-8b36d68df438