canopy VS pre-commit

To create a PineCone account, sign up via this link: https://www.pinecone.io/

So you might have seen in my last blog post that I showed you how to accelerate video frame extraction using GPU's and Babit multimedia framework. In this blog we are going to improve upon our video frame extractor and create a video similarity search(Reverse video search) utlizing different RAG(Retrival Augemented Gerneation) concepts with Pinecone, the vector database that will help us build knowledgeable AI. Pinecone is designed to perform vector searches effectively. You'll see throughout this blog how we extrapulate vectors from videos to make our search work like a charm. With Pinecone, you can quickly find items in a dataset that are most similar to a query vector, making it handy for tasks like recommendation engines, similar item search, or even detecting duplicate content. It's particularly well-suited for machine learning applications where you deal with high-dimensional data and need fast, accurate similarity search capabilities. Reverse video search works like reverse image search but uses a video to find other videos that are alike. Essentially, you use a video to look for matching ones. While handling videos is generally more complex and the accuracy might not be as good as with other models, the use of AI for video tasks is growing. Reverse video search is really good at finding videos that are connected and can make other video applications better. So why would you want to create a video similarity search app?

Pinecone: A scalable vector database service that facilitates efficient similarity search in high-dimensional spaces. Ideal for building real-time applications in AI, such as personalized recommendation engines and content-based retrieval systems.

In this post we’ll build a Python script that uses StripeDocs Reader, a loader on LlamaIndex, that creates vector embeddings of Stripe's documentation in Pinecone. This allows a user to ask questions about Stripe Docs to an LLM, in this case OpenAI, and receive a generated response.

Pinecone is a managed vector database service that simplifies the process of building and scaling vector search applications. It offers a simple API for embedding vector search into applications, providing accurate, scalable similarity search with minimal setup and maintenance.

In case of AIMD, I am doing this all in-memory, but you could also do this in a database (e.g. Pinecone). It all depends on how much data you have and how much compute you have available.

A RAG implementation's quality and performance highly depend on the similarity-based search of embeddings. The challenge arises from the fact that embeddings are usually high-dimensional vectors, and the knowledge base may have many documents. It's not surprising that the popularity of LLM catalyzed the development of specialized vector databases like Pinecone and Weaviate. However, SQL databases are also evolving to meet the new challenge.

Today we are going to look at how to setup Black (a python code formatter) and pre-commit (a package for handling git hooks in python) to automatically format you code on commit.

# See https://pre-commit.com for more information # See https://pre-commit.com/hooks.html for more hooks repos: - repo: https://github.com/pre-commit/pre-commit-hooks rev: v3.2.0 hooks: - id: trailing-whitespace - id: end-of-file-fixer - id: check-yaml - id: check-toml - id: check-added-large-files - repo: local hooks: - id: tox lint name: tox-validation entry: pdm run tox -e test,lint language: system files: ^src\/.+py$|pyproject.toml|^tests\/.+py$ types_or: [python, toml] pass_filenames: false - id: tox docs name: tox-docs language: system entry: pdm run tox -e docs types_or: [python, rst, toml] files: ^src\/.+py$|pyproject.toml|^docs\/ pass_filenames: false - repo: https://github.com/pdm-project/pdm rev: 2.10.4 # a PDM release exposing the hook hooks: - id: pdm-lock-check - repo: https://github.com/jumanjihouse/pre-commit-hooks rev: 3.0.0 hooks: - id: markdownlint

Pre-commit hooks act as the first line of defense in maintaining code quality, seamlessly integrating with linters and code formatters. They automatically execute these tools each time a developer tries to commit code to the repository, ensuring the code adheres to the project's standards. If the hooks detect issues, the commit is paused until the issues are resolved, guaranteeing that only code meeting quality standards makes it into the repository.

Pre-commit Hooks: Pre-commit is a tool that can be set up to enforce coding rules and standards before you commit your changes to your code repository. This ensures that you can't even check in (commit) code that doesn't meet your standards. This allows a code reviewer to focus on the architecture of a change while not wasting time with trivial style nitpicks.

Ah, fair enough!

On my team we use pre-commit[0] a lot. I guess I would define the history to be something like "has this commit ever been run through our pre-commit hooks?". If you rewrite history, you'll (usually) produce commits that have not been through pre-commit (and they've therefore dodged a lot of static checks that might catch code that wasn't working, at that point in time). That gives some manner of objectivity to the "history", although it does depend on each user having their pre-commit hooks activated in their local workspace.

[0]: https://pre-commit.com/

Pre-commit is a framework for managing and maintaining multi-language pre-commit hooks. It supports hooks for various programming languages. Using this framework, you only have to specify a list of hooks you want to run before every commit, and pre-commit handles the installation and execution of those hooks despite your project’s primary language.

You can learn more here: pre-commit.com

Note this includes support for pre-commit.

As a nice next step, you could also add mypy to check your type hints are consistent, and automate running all this via pre-commit hooks set up with… pre-commit.