Contact Me at [email protected] with concerns

TL; DR - Join Slack + Get OpenAI key if you want to run collab notebooks (try this one)

Hi and welcome to Cheat at Search with Agents! To ensure you have a successful class, I want to share some important information to have ready before class. You may have a lot of search foundational knowledge, and might be able to skip this. But for those not 100% confident, here are some prep materials to scan to feel prepared.

You are welcome to come to the class even if you don’t feel 100% confident with skills here, but I want to just set out some expectations.

Please don’t hesitate to get in touch if you’re having a hard time or have any questions. Especially get in touch if you’re uncertain reading this if you’ll get value out of this class!

An OpenAI Key

  1. You’ll want to obtain an OpenAI Key for examples - Visit platform.openai.com to get a key.
  2. You’ll be prompted for it when you run the notebooks to call out to OpenAI
  3. How to get discounted / free OpenAI Key - You can get a discounted OpenAI key (free up to a certain number of tokens) by enabling data sharing. I keep this on for experiments that don’t use customer data (such as these classes). Or if your work has a key you can use, I recommend using this

Join search relevance slack

  1. To communicate about the course, please join this slack community via this link
  2. DM me so I can add you to a private channel for the cohort

Python programming expectations

Here are some basic python programming expectations we’d expect.

  1. Comfort Python programming - comfort programming in Python
  2. Familiarity running a Google colab notebook - all of our examples will be done with Google Colab.
  3. Familiarity with basic Python data stack (pandas, numpy, etc) - many of our examples will use these tools. Pandas is really about working with a “table” aka a “dataframe” sort of in a similar way you’d manipulate a spreadsheet
    1. Numpy for absolute beginners
    2. Pandas in 10 minutes

Search knowledge

It’s a good idea to come to some or all of the sessions for Cheat at Search: Essentials. A free class covering basic search knowledge.

  1. Familiarity with basic search evaluation metrics such as NDCG - We will be experimenting with search datasets as part of the class. I encourage you to ensure you have a solid grasp on metrics such as NDCG. These metrics take query - document labels (which documents, products, etc are relevant results — or not — for a search query). Then taking the results produced by our algorithm to compute a “number” to put on search

    Some pre-reads:

    1. What is a ‘judgment list’?
    2. https://opensourceconnections.com/blog/2019/12/09/demystifying-ndcg-and-err/
    3. Intro to Search Eval / NDCG slidedeck + notebooks
    4. Intro to Search Eval video