Metropolis NIM Workflows

This repository hosts reference workflows that show how to build applications using NVIDIA Inference Microservices (NIMs)

The workflows in this repository require an API key for accessing NIMs. This API Key can be generated at build.nvidia.com. The API Key is used to authorize requests sent to the NIMs and allows access to variety of AI models such as LLMs, VLMs and CV models.

Getting Started

1. Clone Repository

git clone https://github.com/NVIDIA/metropolis-nim-workflows

2. Generate API Key

Go to build.nvidia.com and create an account. Each new account can receive up to 5000 free credits. One request to a NIM uses one credit.

After making an account, you can get an API Key by selecting any of the available NIMs then in the example code section, click on "Get API Key" then "Generate Key".

You will then see your API Key that will look something like "nvapi-xxx-xxxxxxxx_xxxxxxxxxxxxxxxx_xxxxxxxxxxx-xxxxxxx-xxxxxxxxxxxxxx". This full key is what will be used in the example notebooks and scripts.

3. Python Setup

Ensure you have Python 3.10 or above installed on your system. Refer to the Python documentation for download and installation intructions.

4. NIM Workflows

You can now continue and explore the NIM workflows:

• VLM Alert System
• NVCLIP Semantic Search
• Vision Structured Text Extraction
• NVDINOv2 Few Shot Classification