Another year has passed me by. Rather than reflect on the year that passed, I'd like to end the year looking forward into 2025. Hopefully, I 'll get to catch some beautiful equatorial lights in my home country Singapore a little more in the coming year. Meanwhile, in my final post for 2024, I'd like to cram in two Oracle Autonomous Database (ADB) features that deserves more limelight and attention.
ICYMI, earlier this month, Oracle announced the general availability of Meta's Llama 3.2 90B and 11B models on the OCI Generative AI (GenAI) service. As always, Oracle APEX is a great web development platform to give this new multimodal large language model (LLM) a spin. Here's a quick writeup on what I learned over the last weekend.
There a few reasons why we would want to perform "generative AI" tasks, such as generating embeddings, summary, and text, using local resources, rather than relying on external, third-party providers. Perhaps the most important ones are data privacy, costs, and the convenience of choosing what large language models (LLMs) to use. About four months ago, I wrote that with the Oracle Database 23ai 23.5.0.24.06 release on the Oracle Autonomous Database (ADB), it was possible to create the credentials to access these third-party LLMs through their REST APIs. Fast-forward, my ADB now has the upgraded version 23.6.0.24.10 that has support for "local" third-party REST providers, specifically, Ollama. In this article, I will demonstrate how you can utilise other Oracle Cloud Infrastructure (OCI) managed resources to quickly get started.
In previous blog posts, I described how you can create a question and answer application using LangChain and the Oracle Cloud Infrastructure (OCI) Generative AI service. With the Oracle Database 23ai release, we looked at the power of AI Vector Search. This no-cost, built-in database feature provides another alternative to create Retrieval Augmented Generation (RAG) applications using just PL/SQL. However, both solutions require a good amount of code. With the recent availability of the OCI Generative AI Agents service, you can rapidly create a RAG application using low code!
In my previous blog post on Get Started with Oracle Database 23ai AI Vector Search, I mentioned that it was not yet possible to use third-party REST APIs to perform tasks like generating embeddings and text.
This is only the tip of the iceberg. Next up is text generation but there is currently an issue that prohibits the adding of credentials required to work with third-party REST APIs using the
DBMS_VECTORandDBMS_VECTOR_CHAINpackages on the ADB. Hopefully, this will be fixed very soon.
I am happy to post that with the latest 23.5.0.24.06 database release update, this is no longer an issue!
While preparing for my upcoming Kscope24 presentations, I thought I'd introduce a section on Oracle's recently announced AI Vector Search that is now generally available with the Oracle Database 23ai rebranding and release. At the time of writing, you can either provision the new version of the database on the Oracle Cloud Infrastructure (OCI), download the Oracle Autonomous Database (ADB) Free container image, or the Oracle Database 23ai Free. On the OCI, you can either provision an Always Free ADB, or an Oracle Base Database service instance.
The documentation provided by Oracle is comprehensive and should get you started. Here, I condensed everything that I have learned so far into a step-by-step guide to get you started quickly.
Staring from langchain_community version 0.0.31, Oracle introduced a new LangChain document loader that allows you to load the results of an SQL query, against an Oracle Autonomous Database (ADB), as documents that your retrieval-augmented generation (RAG) application can use for text generation tasks. In a previous article, I described how the select ai feature available in ADBs combines the use of database metadata and large-language models (LLMs) to generate an SQL statement that answers a question asked in natural language. In this article, we will take a quick look at how we can load data from the ADB, generate embeddings, then use an LLM to respond to human questions.
Now that I caught your attention, I hope the image isn't too triggering. I grew up learning to fear these little black monsters that potentially carry a load of Dengue virus that could make you really sick. As recently reported, the disease continues to be an issue in Singapore. During my undergraduate years, I was deeply passionate about viral diseases, and spent a few months in the laboratory learning to characterize the structural components of the virus using cyrogenic electron microscopy. Hence, when Luc Demanche challenged me again to demonstrate retrieval-augmented generation (RAG), I decided to collect and use a few recent articles (PMIDs 38140548, 37478848, 37114191), and use them to create a quick proof-of-concept Oracle APEX application that allows the user to ask questions about the disease. These articles cover topics on the Dengue virus, its genomics and ultrastructure, and progress in vaccine development.
In case you missed it, for the last few blogs posts, I had been using Stability AI's Stable Diffusion model to generate the banner images, and I have been mostly successful. Did I really need to? Perhaps not. For this post, I spent almost 30 minutes trying to figure out the "right" prompt to generate the image I wanted: "A car stopped in front a stop sign."
A new year, a new product! Yesterday, Oracle announced the general availability of the OCI Generative AI (GenAI) service. In my final blog post last year, I briefly looked at the beta information that was available to the public, and suggested how you can quite easily add powerful GenAI capability in Oracle APEX applications by calling the REST APIs using APEX_WEB_SERVICE. With the service in GA, I am now finally able to take a deeper dive and hopefully, help you get started.
In the last one year since OpenAI popularised generative AI, a lot of the excitement and intrigue have been centred on the generative power of large language models (LLMs) or foundational models. Understandably so. However, the generative power of this models only cover a segment of business needs. The underlying promise of these frontier models is in its ability to "understand" natural language, and that to me, is where we will find more utility.
For the last few years at Insum, we've had an almost biannual tradition of reviewing what's been going on in the AI/ML space, in particular, around what Oracle has been doing to provide innovative solutions and tools. If you missed those, I have provided the links below. In the absence of an updated review, I am hoping that this final 2023 blog post would serve somewhat as a stopgap.
In my previous post, you read about how I performed fine-tuning and deployed a large language model (LLM) on the Oracle Cloud Infrastructure (OCI) Data Science service. Here, I was hoping to do the same with the Stable Diffusion XL, a Latent Diffusion Model (LDM), but unfortunately, the platform currently does not support a Hugging Face pipeline using this model. I found an alternative, and you will read about it later in this article.
To demonstrate its utility, I create a simple Oracle APEX application that reads NFC tags using the experimental Web NFC API, and then displays an image generated by the deployed machine learning (ML) model.
One of the key highlights at Oracle Cloudworld 2023 is general availability of Select AI feature that allows anyone to query the Oracle Database using natural language. This is likely made possible with a template wrapping the database metadata with the prompt, and then calling either the OpenAI or Cohere APIs to generate the intended text.