This article walks you through the process of creating robust prompts for LLMs. Learn how slight variations in your prompt can shape your model response. — What is Prompt Engineering? Prompt Engineering refers to crafting effective prompts that can efficiently instruct LLMs that power Bard or ChatGPT to perform desired tasks. Those tasks can be applied to many different use cases. It involves the precise wording, formatting, and structuring of a prompt to optimize the model’s output, enhancing relevance and…

A comprehensive list of article that provides you knowledge about Machine Learning on Google Cloud. — This website is for the sole purpose of keeping an overview of all ML related articles I published on medium.com. Almost every article contains a notebook with code for you to use including a YouTube video. Generative AI Generative AI — Getting Started with PaLM 2 Generative AI — The Evolution of…

Learn how to get started with Generative AI and PaLm2 on Vertex AI — The Vertex AI PaLM API is a Google Cloud service that enables the usage and training of generative models. Let me first remove some of the confusion. Generative AI Studio is a feature of Vertex AI. …

How to Approach ML Problems in Times of Generative AI. The world of ML is rapidly changing. — With the advent of large language models that power Generative AI such as PaLM by Google and GPT by OpenAI, the recent landscape has considerably shifted. These large LLMs that can only be trained by companies with significant budgets have begun to alter the approach of ML engineers toward solving…

Say goodbye to unexpected costs and unnecessary running notebooks — Vertex AI provides two types of notebooks: Google-managed and user-managed. Google-managed notebooks automatically shut down after a period of inactivity, which helps save costs and prevent billing surprises. However, there isn’t an integrated method for automatically shutting down user-managed notebooks — until now. Thanks to Jose Brache, you can now…

Google’s Vertex AI Notebooks are currently running on Python 3.7. You might need a more recent Python version to use newer Python features. — Fortunately, we can install a newer version in addition to the existing Python version using Conda. This way you can choose both kernels in the Vertex AI Workbench Notebook UI. Open a terminal in the notebook and create a new conda environment with the Python version you need.

A production scale vector similarity search with only 100 lines of code — Implementing the infrastructure to dynamically update and query a similarity index like Facebook’s Faiss or Spotify’s Annoy is a huge implementation effort. With the architecture in this article, we get it into production in minutes. If you haven't worked in that topic space before, I can recommend checking out my…

Batch predictions allow you to predict large amounts of data in parallel. — Vertex AI Batch Prediction is made for large datasets that would take too much time with an online prediction approach. It provides a scalable, serverless, and efficient service for cases where you don’t need an immediate response (asynchronous). Jump to the Notebook and Code All the code for this article is ready to use in a…

A simple way to circumvent the limited debugging and logging capabilities. — Vertex AI Batch Prediction jobs do not log errors to Cloud Logging. This is mainly due to the batch job running in a Google-managed project. This can be a big issue if something is not working as expected. Without logging, we are running nearly blind. Batch Predictions in production are…