Application Examples#
End-to-end applications built on Nemotron models, including agentic workflows, RAG systems, and fine-tuning pipelines. Each card links to its directory in the Nemotron GitHub repository.
Introductory notebook covering Nemotron 3 Super’s reasoning features: thinking, reasoning budget, low effort mode, streaming responses, tool-call streaming, and Perplexity Search integration using the OpenAI-compatible API.
End-to-end LoRA fine-tuning of Nemotron 3 Nano on Text2SQL (BIRD SQL) with deployment via NVIDIA NIM or vLLM using NeMo AutoModel or Megatron Bridge.
IDP pipeline that extracts and queries complex enterprise documents — financial reports, charts, and tables — using NeMo Retriever and multimodal Nemotron models.
End-to-end voice-driven RAG agent combining speech-to-text, multimodal retrieval, 1M-token reasoning, and safety guardrails using open Nemotron models.
Introductory notebook covering basic inference, reasoning mode toggling, and multi-agent systems using the OpenAI-compatible API via OpenRouter and LangChain.
Natural language-driven ML agent built on Nemotron Nano 9B with GPU-accelerated data exploration and model training using RAPIDS cuDF and cuML.
Production-ready RAG agent using local Hugging Face embedding and reranking models with NVIDIA AI Endpoints for LLM inference, built on LangGraph.