nemotron-super3
OfficialAnswer Super3 training and recipe questions fast.
Education & Research#nemotron#quantization#model architecture#super3#training stages#evaluation benchmarks#recipe mapping
AuthorNVIDIA-NeMo
Version1.0.0
Installs0
System Documentation
What problem does it solve?
This Skill removes the need to manually search across scattered papers, model-card notes, and recipe docs when you want facts about NVIDIA Nemotron 3 Super (Super3).
Core Features & Use Cases
- Super3 knowledge desk: answers architecture, training stages (pretrain/SFT/RL/eval/quantization), evaluation setup, and deployment-oriented checkpoint differences.
- Paper vs. open-repo clarity: consistently distinguishes reported research claims from what the released recipes can reproduce.
- Routing-first retrieval: directs answers through the skill’s internal map (INDEX → quick reference → smallest relevant file) to ensure citations and scope correctness.
Quick Start
Ask the Skill how LatentMoE and MTP work in Nemotron 3 Super by invoking /nemotron-super3 and requesting an architecture explanation with the relevant paper sections.
Dependency Matrix
Required Modules
None requiredComponents
Standard package💻 Claude Code Installation
Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.
Please help me install this Skill: Name: nemotron-super3 Download link: https://github.com/NVIDIA-NeMo/Nemotron/archive/main.zip#nemotron-super3 Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
Agent Skills Search Helper
Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.