📚 Collections of common [application patterns], [tools], [models], and [research papers] for building AI applications. 🔍
🔥 New applications and content updated every week! 📅
Have ideas for other patterns? Open an PR and create a new folder under examples. Or simply open an issue and we can discuss it.
| Name | Example Description | Community Projects |
|---|---|---|
| LLM Agent | Thought, Action, Observation Agent built from scratch | Microsoft's Jarvis, Google's Palm, AutoGPT |
| ChatGPT Plugin | A plugin looking up real time Bart schedule | |
| Deploy Self-Hosted LLM | Running your own LLM on Replicate | |
| Semantic Search and Q&A Bots | 🚧 Under Construction |
Common tools you will come across as you explore projects built in the community.
| Name | Category | Description |
|---|---|---|
| Flask | Framework | A lightweight web framework for Python. Since Python is the language of AI, you might as well use a web framework in Python. Another common option is FastAPI. |
| Next.js | Framework | A popular Javascript framework, often used with Vercel. |
| Vercel | Infra & Hosting | A cloud platform specialized in hosting Next.js apps. |
| Tailwindcss | Framework | A utility-first CSS framework, which means you will define styling directly with pre-built css classes. |
| Pinecone | Infra & Hosting | A vector database for AI applications. This is commonly used to store embeddings of data and enable similarity based search (More on this in Q&A bots). Alternatively, you can also use Milvus, Faiss, Chroma and Weaviate and others. |
| Supabase | Infra & Hosting | A popular PostgresQL as a service vendor. Works well with NextJS and offers Authentication and storage APIs out of the box. Also has support for storing embeddings (using a Postgres plugin - pgvector) and files. |
| LlamaIndex | Framework | A library that provides an interface between LLMs and external data by handling data connectors, building indices and querying (e.g., including managing prompt token size, etc.). |
| LangChain | Framework | A library that helps you build LLM based complex action chains and agents. |
| OpenAI API | API | Currently the state of the art models. Commonly used APIs include:
|
| HuggingFace | Infra & Hosting | The GitHub for AI models, a platform to build, deploy, and share models across the community. An alternative to OpenAI models. |
| ElevenLabs | API | AI text to speech API. Voice cloning. |
| Replicate | API | Run open-source machine learning models with a cloud API. |
A list of the more "hackable" open source models and how to work with them.
| Name | Description | Size | Data | OSS & License | How to work with it |
|---|---|---|---|---|---|
| LLaMA | Foundational 65B LLM from Meta AI | 65B | Yes (Research only) | Hosting | |
| Dolly 2.0 | Open source, instruction-following LLM | 12B | 15,000 high-quality human-generated prompt / response pairs | Yes (Commercial use friendly) |
🚧 Under Construction