OpenAI's Evals Framework, Tabyl, LangChain’s Cypher Search, Generative AI with ChatGPT and OpenAI Models

A/B Optimization with Policy Gradient Reinforcement Learning                                                                                                                                                                                                                                                                                                                                                                                                                 [Open in app]( or [online]() [OpenAI's Evals Framework, Tabyl, LangChain’s Cypher Search, Generative AI with ChatGPT and OpenAI Models]( A/B Optimization with Policy Gradient Reinforcement Learning May 26   [Share](   Brought to you by... Connect. Configure. Control. See how Drata simplifies compliance management. [Drata]( [Drata]( evidence collection with 80+ integrations and monitors risk 24/7 for 14+ frameworks. Whether it’s SOC 2, ISO 27001, GDPR, or HIPAA, you can stay compliant without the messy, manual work. Book a demo and see why companies like Notion and Lemonade choose Drata to streamline their compliance programs.   Plus, [Packt]( get 10% off and waived implementation fees.  [REQUEST A DEMO]( --------------------------------------------------------------- 👋 Hey, "Machine learning algorithms will become our digital co-pilots, augmenting our abilities and transforming industries." - [Peter Norvig, Computer Scientist.]( Embracing the digital co-pilot means harnessing the power of AI and machine learning algorithms to augment our abilities and transform industries. By adopting a mindset of collaboration, we can unlock limitless potential and achieve unprecedented feats. With AI as our trusted companion, we gain access to vast data analysis, pattern recognition, and predictive insights. It highlights our resilience, adaptability, and unwavering commitment to innovation. Also, it is a paradigm shift that propels us toward a future of boundless possibilities.  Now, let's shift our focus to the featured resources of this week in [DataPro#45]( which delve into ways of embracing AI capabilities in our professional practice to stay ahead in today's rapidly evolving technological landscape. "[Modern Generative AI with ChatGPT and OpenAI Models]( the book we are releasing today, will be the cornerstone of this endeavor. Additionally, we have an exciting lineup of topics that revolve around optimizing the utilization of [OpenAI's Evals Framework]( improving the operational efficiencies of [Apache Iceberg tables]( exploring LangChain's [Cypher Search]( and examining Auditoria.AI's development of [AI-powered smart assistants](. Furthermore, we will gain insights into the governance of [superintelligence]( apply A/B optimization with Policy [Gradient Reinforcement Learning]( and utilize [Tabyl]( a modern frequency table designed for R users.  Whether you are an experienced data professional, or someone interested in utilizing generative Al to enhance productivity, these resources offer the perfect solution to help you achieve your goals. Prepare yourself for a productive and enlightening learning experience!  If you’re interested in sharing ideas to foster the growth of the data community, then this survey is for you. Consider sharing your thoughts and get a FREE bestselling Packt ebook as PDF. Jump on in!  [TELL US WHAT YOU THINK]( Cheers, Merlyn Shelley Editor in Chief, Packt This Week’s Key Highlights: - [Benchmarking for AlloyDB and how to do it]( - [Real-time time series anomaly detection for streaming application]( - [How well do GPT models follow Prompts]( - [The Notorious XGBoost]( - [Improve your Gradient Descent: The Epic Quest for the Optimal Stride]( Latest Research on GitHub - [microsoft]( Composable Diffusion (CoDi) generates diverse outputs by combining various input modalities like language, image, video, or audio.  - [kyegomez]( Tree of Thoughts implementation enhances large language models' reasoning by over 70% for deliberate problem-solving.  - [aiwaves-cn]( RecurrentGPT simulates LSTM recurrence with natural language paragraphs, replacing vectorized elements and utilizing prompt engineering.  - [microsoft]( Guidance empowers efficient and effective control of modern language models, surpassing traditional prompting or chaining methods.  - [facebookresearch]( Fairseq(-py) is a versatile toolkit for training custom models in sequence modeling for translation, summarization, and language tasks.  - [microsoft]( Pengi utilizes transfer learning by treating audio tasks as text-generation tasks, making it an audio language model.  [Pledge your support]( Industry Insights AWS ML - [Improve operational efficiencies of Apache Iceberg tables built on Amazon S3 data lakes:](  This post explores operational use cases for [Apache Iceberg]( tables on an [Amazon S3 data lake]( and [Amazon EMR](. It highlights important aspects like storage cost optimization, disaster recovery, multi-region access, and handling increased request rates. The focus is on improving operational efficiencies for Apache Iceberg tables.  - [Real-time time series anomaly detection for streaming applications on Amazon Kinesis Data Analytics:]( This post introduces a streaming time series anomaly detection algorithm using [Apache Flink]( and [Amazon Kinesis Data Analytics](. It explains the algorithm based on matrix profiles and left-discords, provides a working example, and highlights the potential of Kinesis Data Analytics Studio for real-time data visualization. Further implementation details can be found in the accompanying [GitHub repository]( Google AI & ML - [Benchmarking for AlloyDB and how to do it:]( This blog post discusses the AlloyDB cache and its accessibility by different processes. It mentions a separate [blog]( by Ravi Murthy that delves into the storage layer of AlloyDB. The article highlights how benchmark test settings can impact the relevance of results and explores the question of the right testing approach, which depends on factors such as using real data and workload or comparing database deployments. While there is no universal solution, certain approaches can be sensible.  - [How Auditoria.AI is building AI-powered smart assistants for finance teams:]( Auditoria.AI aims to [automate]( routine tasks in finance and accounting using AI and natural language technology. By eliminating mundane tasks like data entry and document validation, finance professionals can focus on providing strategic insights. They leverage Google Cloud's [Document AI]( to extract relevant information from various document types, improving efficiency and accuracy.  ChatGPT Expert Insights from Packt Community  Domains of generative AI – By [Valentina Alto]( AI has been making significant strides in recent years, and one of the areas that has seen considerable growth is generative AI. Generative AI is a subfield of AI and DL that focuses on generating new content, such as images, text, music, and video, by using algorithms and models that have been trained on existing data using ML techniques.  [Generative AI]( The fact that generative AI is used in many domains also implies that its models can deal with different kinds of data, from natural language to audio or images. Let us understand how generative AI models address different types of data and domains.  Text generation  One of the greatest applications of generative AI—and the one we are going to cover the most throughout this book—is its capability to produce new content in natural language. Indeed, generative AI algorithms can be used to generate new text, such as articles, poetry, and product descriptions.  Image generation  One of the earliest and most well-known examples of generative AI in image synthesis is the Generative Adversarial Network (GAN) architecture introduced in the 2014 paper by I. Goodfellow et al., Generative Adversarial Networks. The purpose of GANs is to generate realistic images that are indistinguishable from real images. This capability had several interesting business applications, such as generating synthetic datasets for training computer vision models, generating realistic product images, and generating realistic images for virtual reality and augmented reality applications. You can [read the full chapter for free here...]( This excerpt is taken from the recently published book titled "[Modern Generative AI with ChatGPT and OpenAI Models]( written by [Valentina Alto]( and released today, May 2023. To get a preview of the book's content, be sure to read the [whole chapter available here]( or sign up for a [7-day free trial]( to access the complete Packt digital library. To explore more, click on the button below.  [DISCOVER FRESH CONCEPTS & KEEP READING!]( Latest Developments in LLMs & GPTs - [Governance of superintelligence:]( This post discusses the potential benefits and risks associated with superintelligence. While superintelligence offers the possibility of a prosperous future, careful risk management is necessary. Mitigating the risks of current AI technology is important, but dealing with superintelligence requires special treatment and coordination. Stopping its creation is challenging due to decreasing costs, increasing actors involved, and its inherent role in technological progress. Thus, getting the approach right is crucial.  - [LangChain has added Cypher Search:]( This blog post introduces the LangChain library, which facilitates the generation of Cypher queries for efficient data retrieval from Neo4j. It demonstrates how to utilize the newly added Cypher Search feature. The post also highlights the responsiveness of LangChain's maintainers and their openness to new ideas. The code can be found on [GitHub]( - [How well do GPT models follow Prompts:]( Prompt engineering plays a crucial role in maximizing the output potential of large language models (LLMs). Users can modify prompts to achieve desired results from LLMs. This article guides machine learning engineers in evaluating prompt effectiveness using OpenAI's GPT-3.5 Turbo model for text summarization of news articles.  Find Out What’s New - [A/B Optimization with Policy Gradient Reinforcement Learning:]( This article explores the use of policy gradient reinforcement learning for A/B optimization. It provides a simple demonstration of the policy gradient method, explaining its underlying mechanism and displaying the learning process. The demo illustrates how a neural network agent optimizes its policy by maximizing rewards. The policy gradient method, like backpropagation, updates the network's parameters. The article emphasizes the application of reinforcement learning in A/B optimization and highlights the challenge of selecting the appropriate learning rate. A code and [streamlit demo can be found on huggingface.co](.   - [How To Best Leverage OpenAI’s Evals Framework:]( A recent [survey]( reveals that 43% of data science and engineering teams plan to deploy large language models (LLMs) in the coming year. Evaluating LLMs is challenging, but OpenAI has released their [Eval Framework]( a tool to compare LLM performance against benchmarks and other models. OpenEvals offers standardized metrics and tasks, making it valuable for evaluating and comparing LLMs as they advance.  - [From Data Engineering to Prompt Engineering:]( This article explores the use of ChatGPT and Python to solve data engineering tasks, such as data ingestion, transformation, and quality assurance. It highlights the connection between data engineering and prompt engineering. ChatGPT successfully implemented prompts and even corrected mistakes, but testing, refactoring, and optimization are still necessary. The future of data engineering may involve a shift from coding to prompt engineering.  - [The Notorious XGBoost:]( is a highly effective algorithm for supervised learning tasks, known for its ability to capture complex non-linear relationships accurately. This article highlights the mathematical foundations behind XGBoost's success and its reliability. While the future impact of more advanced models remains uncertain, understanding XGBoost's power is valuable. It emphasizes the importance of ethical principles and best practices in using machine learning techniques for equitable and beneficial outcomes.  - [Tabyl: A Frequency Table for the Modern R User:]( This article introduces the tabyl function in R's janitor package for creating frequency tables. Through practical examples, it showcases the user-friendly and flexible nature of tabyl(), which excels in computing proportions and generating tidy data frames that integrate well with the tidyverse ecosystem. The function's well-structured outputs, along with the ability to enhance them using adorn functions, make it a valuable tool for data analysis in R.  - [Improve your Gradient Descent: The Epic Quest for the Optimal Stride:]( Gradient Descent is a widely used technique for optimizing machine learning models. This article explores three techniques to optimize the step size in gradient descent: Fixed Step Size, Exact Line Search, and Armijo Backtracking. While these are fundamental techniques, there are numerous other methods available. The article emphasizes the importance of understanding optimization intricacies for effective algorithm tuning.  See you next time! As a GDPR-compliant company, we want you to know why you’re getting this email. The _datapro team, as a part of Packt Publishing, believes that you have a legitimate interest in our newsletter and its products. Our research shows that you opted-in for email communication with Packt Publishing in the past and we think your previous interest warrants our appropriate communication. If you do not feel that you should have received this or are no longer interested in _datapro, you can opt out of our emails by clicking the link below.   [Like]( [Comment]( [Restack](   © 2023 Copyright © 2022 Packt Publishing, All rights reserved. Our mailing address is:, Packt Publishing Livery Place, 35 Livery Street, Birmingham, West Midlands B3 2PB United Kingdom [Unsubscribe]() [Start writing]()