GPT Neo on E2E Cloud, Hugging Face Agents, Facies Classification, U-Net in PyTorch, Bayesian models

Practical Guide to Azure Cognitive Services                                                                                                                                                                                                                                                                                                                                                                                                                 [Open in app]( or [online]() [GPT Neo on E2E Cloud, Hugging Face Agents, Facies Classification, U-Net in PyTorch, Bayesian models]( Practical Guide to Azure Cognitive Services May 19   [Share](   👋 Hey, "MLOps is about establishing a culture of collaboration and continuous improvement, where data scientists, engineers, and operations work together to deliver reliable and scalable machine learning solutions." - [David Aronchick, Head of Compute over Data, Protocol Lab]( Unleashing the full potential of machine learning relies on seamless collaboration and continuous improvement. In MLOps, data scientists, engineers, and operations professionals join forces, combining their expertise to foster an innovative data culture. Together, they create reliable and scalable machine learning solutions, driving organizations to new heights. This week in [DataPro #44]( we have an exciting lineup of topics to explore. We'll dive into [Automatic Prompt Optimization]( discover the capabilities of [Hugging Face Agents]( and learn about AI-powered code suggestions, as well as security scans, within [Amazon SageMaker notebooks]( But that's not all! We have a whole chapter unlocked from our recently published [Practical Guide to Azure Cognitive Services]( where we delve into a business case study on solving operational challenges. Furthermore, we'll explore useful tools like deploying [GPT Neo on E2E Cloud]( cooking your first [U-Net in PyTorch]( demystifying [Bayesian models]( and uncovering six hidden features of [Pandas]( ready for a productive and insightful 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 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: - [Facies Classification using Snowflake Integration]( - [AI-powered code suggestions and security scans]( - [Optimizing Vector Quantization Methods]( - [Enabling conversational interaction on mobile with LLMs]( Sponsored Content This Microsoft Build, reach your potential with a Microsoft Learn Cloud Skills Challenge  Sharpen, expand, and discover new skills by completing any one of eight unique collections on Microsoft Learn. Successful completion can earn you a free Microsoft certification exam! Sign up now for any one of the following topics:  - Azure Cosmos DB Developer  - .NET  - Azure AI  - Developer Tools  - Cloud Development  - DevOps   - Microsoft 365 Developer  - Power Platform Developer  The challenge begins on May 23rd when Microsoft Build kicks-off and finishes on June 20th. [Register now]( [REGISTER NOW!]( Latest Research on GitHub - [facebookresearch]( Explains [Decentralized and Accelerated Bundle Adjustment (DABA)]( for tackling bottlenecks in bundle adjustment problems.  - [salesforce]( An AutoML library written in Scala that runs on top of Apache Spark.   - [salesforce]( Official release of CodeT5 and CodeT5+ models for code understanding and generation. - [salesforce]( Library for rapid development and benchmarking of multimodal models.  -  [facebookresearch]( Creates a joint embedding across six modalities and enables diverse applications like cross-modal retrieval and generation.  - [deepset-ai]( Open-source NLP framework for interacting with data using Transformer models.  [Pledge your support]( Industry Insights AWS ML - [Facies Classification using Snowflake Integration and train using SageMaker Canvas]( Facies classification involves identifying lithologic formations from wellbore data. Geologists analyze depth-dependent information obtained from wireline logs to determine potential facies. AI and machine learning are increasingly used for this task, but many oil companies lack the necessary expertise. This post explains how to import data from [Snowflake]( to [Amazon SageMaker Canvas]( for training a facies classification model.  - [GPT-NeoXT-Chat-Base-20B foundation model for chatbot applications is now available on Amazon SageMaker:]( This post explains how to deploy the [GPT-NeoXT-Chat-Base-20B]( model and utilize it in an [OpenChatKit]( interactive shell. It provides an open source chatbot model for integration into applications. JumpStart models [leverage]( Deep Java Serving with Deep Java Library (DJL) and deep speed libraries to optimize performance and ensure network isolation for improved security.  - [AI-powered code suggestions and security scans in Amazon SageMaker notebooks using Amazon CodeWhisperer and Amazon CodeGuru:]( Amazon SageMaker provides two options for managed notebooks: fast start collaborative notebooks within [Amazon SageMaker Studio]( and [Amazon SageMaker notebook instances]( offering more control over configurations. The newly available extensions, [Amazon CodeWhisperer]( and [Amazon CodeGuru Security]( provide AI-powered code suggestions and ensure code security and AWS best practices. This post demonstrates getting started with these extensions in both Studio and SageMaker notebook instances.  Google AI & ML - [Enabling Conversational Interaction on Mobile with LLMs:]( The paper investigates the application of large language models (LLMs) for [language-based interactions with mobile UIs](. It demonstrates the adaptability of pre-trained LLMs like [PaLM]( and proposes prompting techniques for prototyping and testing language interactions. A novel algorithm for generating text representations of mobile UIs is also introduced. Results indicate competitive performance with minimal data examples per task, showcasing the potential of LLMs in conversational interaction design workflows.  Microsoft Research - [Automatic Prompt Optimization with “Gradient Descent” and Beam Search:]( Automatic Prompt Optimization (APO) is introduced as a method to enhance prompts for Large Language Models (LLMs) using training data and an LLM API. APO uses numerical gradient descent inspired techniques to adjust prompts based on data minibatches, guided by beam search and bandit selection. Preliminary results show that APO outperforms prior prompt editing techniques, achieving up to a 31% performance improvement in NLP tasks.  Understanding Core Concepts Using Anomaly Detector for Discovering Abnormalities – By [Chris Seferlis]( , [Christopher Nellis]( [Andy Roberts]( The Anomaly Detector service monitors all the data being collected and calls out areas of concern to the procurement team, giving them the ability to accept the anomaly and provide feedback to the system. Again, this will improve an ML model that has been developed over time or allow the team to go in a different direction with the transaction. Now let's look at the steps involved in the process, - Data is ingested from some source in Azure.  - Some compute—in this case, Azure Databricks—is used to manipulate and prepare the data and serve it to the Anomaly Detector service for monitoring.  - When anomalous behavior is detected, an event is triggered and sent to Azure Service Bus for further action.  - An alert is sent to appropriate parties for follow-up action using the service of the developer's choosing.  - The output of the data is then logged in Azure Data Lake Storage (ADLS).  - Power BI is used to visualize the results of the logged data.  The above content is extracted from the newly released book, "[Practical Guide to Azure Cognitive Services]( authored by By [Chris Seferlis]( [Christopher Nellis]( [Andy Roberts]( and published in May 2023. To get a glimpse of the book's contents, make sure to read the [whole chapter provided here]( or if you want to unlock the full Packt digital library free for 7 days, [try signing up now]( To learn more, click on the button below.  [DISCOVER FRESH CONCEPTS & KEEP READING!]( Find Out What’s New - [Deploying GPT Neo on E2E Cloud:]( Data scientists can maximize the potential of GPT-Neo by using E2E Cloud, benefiting from its cost-effectiveness and industry-vetted performance. Learn about GPT-Neo and datasets and follow this tutorial to deploy the open-source language model with the power of NVIDIA GPUs and flexible pricing options.   - [Optimizing Vector Quantization Methods by Machine Learning Algorithms:]( Vector Quantization (VQ) is a data compression technique similar to k-means algorithm. It can model any data distribution by optimizing codebooks. Variants like Residual VQ, Additive VQ, and Product VQ have been optimized using machine learning optimization techniques, such as Noise Substitution in Vector Quantization (NSVQ). This open-source implementation helps choose the best VQ method for specific use-cases, considering trade-offs between bitrate, accuracy, and complexity.  - [Get to know about Hugging Face Agents:]( Hugging Face recently integrated [tools]( and agents into its [Transformers]( library, offering convenience and speed for both technical and non-technical users. Agents are language models that perform tasks based on prompts, while tools are selected by agents to assist in executing the code. Technical users can modify and run the code returned by agents, while non-technical users can simply express their needs in plain English. The combination of agents and tools proves to be powerful and beneficial for a wide range of users.  - [6 Things that you probably didn’t know you could do with Pandas:]( Pandas has become an essential tool for data scientists and analysts due to its powerful functionalities. With over [3 million daily downloads]( Pandas demonstrates its widespread popularity. Despite its fundamental features, there are hidden gems that can enhance data analysis. This blog highlights six interesting things to supercharge data analysis with Pandas. With its powerful and flexible functionalities, Pandas has become an indispensable tool for data scientists and analysts.  - [Cook your First U-Net in PyTorch:]( U-Net is a deep learning architecture used for image segmentation tasks. It produces high-quality segmentation masks with sharp boundaries and is widely adopted in various domains. This tutorial explores [U-Net]( its effectiveness in biomedical and other image segmentation tasks, and provides a PyTorch implementation guide.  - [Demystifying Bayesian Models: Unveiling Explanability through SHAP Values:]( This article explains a Bayesian model using the SHAP framework and PyMC, a probabilistic programming library for Python. The approach involves applying SHAP to an ensemble of deterministic models generated from a Bayesian network, providing explainability through SHAP value samples. The article provides a simple example and accompanying [notebook]( for implementation.  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]()