GenAI and LLMs development, trends and implications (7. - 13.10.2024)

Adoption:

• Bursting the Gen AI hype bubble - once and for all
• Basics - 10 critical AI concepts explained in 5 minutes
• Generally AI - Fantastic algorithms and where to find them
• Seven factors driving AI adoption
• Exploring the GenAI technology stack and how it all fits together
• Meta unveils Movie Gen, a new AI model for video generation

LLMs generally:

• Chain of Thought prompting for LLMs
• Operationalize a scalable AI with LLMOps principles and best practices
• Prompt Caching in LLMs: Intuition - how caching works in attention-based models
• Improving model quality without building larger models
• PayPal adds GenAI support with LLMs to its Cosmos.AI MLOps platform
• Fine-tuning Llama 3.2 for performance on targeted domains

New models and functionality:

• OpenAI Developer Day 2024 (SF) announces real-time API, vision fine-tuning, and more
• Meta releases Llama 3.2 with vision, voice, and open customizable models
• Claude AI - unboxing Anthropic’s LLM-based AI assistant, artifacts & use cases
• A Walkthrough of Nvidia’s latest multi-modal LLM family

AI agents:

• All about AI agents -  autonomy, reasoning, alignment, and more

AI-enhanced software development:

• Test-Driven Generation - adopting TDD with GenAI

Enhancing applications with GenAI:

• Latest innovations in recommendation systems with LLMs
• DocAI - chatting with AI and asking questions in the context of your documents

IT links (7. - 13.10.2024)

• Reducing memory footprint with Java Object Layout (JOL) library
• Object equality in Java - a degree of equality and fuzzy comparison

       Java Streams:

• Exploring findAny() and anyMatch()
• Effective techniques for optimizing Java Streams - scenarios of when to use Streams, and when to avoid them

• Mocking @Value in Spring Boot test
• Test-Driven Generation - adopting TDD with GenAI

• Multiple ways how to change field value before Update and Insert in Hibernate

• Secret management and rotation in AWS
• Microservice proliferation - too many microservices - the pitfalls associated with the excessive creation of microservices, insights on their causes, implications, and potential strategies for mitigation

• Data modeling and techniques and best practices for optimization in Snowflake
• Data Teams Survey - lag in DataOps and value delivered
• Setting up a Data Mesh organization
• Scaling Uber’s batch data platform - a journey to the cloud with Data Mesh principles

GenAI and LLMs development, trends and implications (30.9. - 6.10.2024)

Adoption:

• Generative AI will require 80% of engineering workforce to upskill through 2027
• Demand is growing for robust computing - hyperscale data centers ignite Construction Surge
• Knowledge-as-a-service - the future of community business models
• Generally AI - generative AI and creativity
• When generative AI isn’t always the answer

LLMs generally:

• What to consider when adopting Large Language Models
• Choosing the architecture for your GenAI application
• Evaluating performance of LLM-based applications
• Do we really need more complex models?

AI agents:

• What makes a true AI agent? Rethinking the pursuit of autonomy
• The intersection of tool calling and reasoning in Generative AI

RAG:

• Chunking strategies - why, when, and how to chunk for enhanced RAG
• Improving RAG quality with knowledge graphs
• Stop guessing and measure your RAG system to drive real improvements
• Using HyDE technique for better LLM RAG retrieval
• Build a RAG-powered chatbot With Gemini, MyScaleDB

AI-enhanced software development:

• The AI-native developers who build software through prompt engineering
• Using Spring AI with LLMs to generate Java tests
• Legacy codebases modernization meets GenAI
• The state of AI completion in JetBrains IDEs - complete the un-completable
• PyCharm - prompting AI directly in the editor
• Comparison of AI coding assistants - Codeium versus Copilot

Enhancing applications with GenAI:

• Spring Petclinic - Implementing an AI assistant with Spring AI
• Supercharging your AI applications with Spring AI Advisors
• ChatClient Fluent API in Spring AI
• OpenAI releases stable version of .NET Library with GPT-4o Support

IT links (30.9. - 6.10.2024)

Java:

• Using Lombok library with JDK 23
• Asserting collection of JSON objects ignoring order

    Reminding basics - now Java regarding exceptions:

• What is exception propagation in Java? Understanding the mechanism and its practical implications
• Understanding the difference between errors and exceptions in Java

Other:

• Modular monoliths using Spring Modulith
• Balance trade-offs in Microservices Architecture

• The Instability-Abstractness-Relationship — An alternative view
• Martin Fowler reflects on refactoring - improving the design of existing code
• Introduction to code analysis tools in IntelliJ IDEA

• How to make technical debt your friend
• Agile primitives - ten core principles

• Stream processing in the serverless world
• 10 years of Kubernetes - its journey, impact, and future

AWS:

• Release of Amazon RDS for MySQL Zero-ETL integration with Amazon Redshift
• AWS is sunsetting more services, including AWS App Mesh and Amazon FSx File Gateway

Začínám psát články v češtině na Seznam Médium

Nedávno jsem se rozhodl začít psát články v češtině na platformě Seznam Médium. Snažím se zaměřit hlavně na vědu a technologie, ale samozřejmě záběr může být o dost větší. Zde je seznam do teď mnou napsaných článků:

Proč se nedožíváme 200 let? Možná za to mohou dinosauři

Rok 2023 v letech do vesmíru

Multivitamíny – klíč k zachování kognitivního zdraví ve stáří?

Vorvani žijí v klanech podobných těm, v jakých žili první lidé

T-lymfocyty jako pramen mládí?

Strojové učení pomáhá překonat jeden z největších problémů jaderné fúze

Aktuální seznam mnou publikovaných článků můžete nalézt na mém profilu zde: Primus Immortails

Exploring Advanced AI Techniques: Ghost Attention, Thought Structures, Prompt Engineering and more

Diving deeper into the realm of generative AI, I've come across several articles that I find interesting as a beginner in this field:

The article - Understanding Ghost Attention in LLaMa 2 - delves deep into the technique of the ghost attention technique in LLaMa 2.

One example of providing instructions for specific chat is Prompt Instructions in Watsonx IBM service:

You define instructions in the upper input and then start to chat below.

One detailed look into how generative AI works is this article exploring the differences between "Chain of thoughts" and "Tree of thoughts" - Chain of Thoughts vs Tree of Thoughts for Language Learning Models (LLMs)

How to work better with these systems? You can improve the output using prompt patterns or n-shot prompting - 7 Prompt Patterns You Should Know

For controlling grounding data used by an LLM and constraining it for your enterprise Gen AI solutions, consider using Retrieval Augmented Generation (RAG). You can see how to use it, for example in Azure, here - Retrieval Augmented Generation (RAG) in Azure AI Search

Additionally, to gain more from LLMs, you can explore architecture patterns and mental models as described here - Generative AI Design Patterns: A Comprehensive Guide

A comprehensive overview of generative AI and LLMs' trends, use cases, and future implications II. - Engineering and development insights

7 weeks (from 4.9. to 22.10.2023) in the world of Large Language Models and Generative AI tools, this time more focused on the engineering side:

Prompt engineering:

Parallel processing in prompt engineering: the skeleton-of-thought technique.

Unlocking reliable generations through Chain-of-Verification - a leap in prompt engineering.

LLMOps: production prompt engineering patterns with Hamilton.

Crafting different types of program simulation prompts - defining the new program simulation prompt framework.

Some kick-ass prompt engineering techniques to boost our LLM models.

And other prompt engineering tips, a neural network how-to, and recent must-reads.

AI Development and Engineering:

The team behind GitHub Copilot shares its lessons from building the app.

Amazon Bedrock for building and scaling generative applications is now generally available.

Experience from building generative AI apps on Amazon Web Services, using Amazon Bedrock and SageMaker.

A guide with 7 steps for mastering LLMs.

Key tools for enhancing Generative AI in Data Lake Houses.

An introduction to loading Large Language models.

Introduction to ML engineering and LLMOps with OpenAI and LangChain.

MLOps and LLM deployment strategies for software engineers.

Modern MLOps platform for Generative AI.

Leveraging the power of LLMs to guide AutoML hyperparameter searches.

LLMs demand Observability-Driven Development.

LLM monitoring and observability — a summary of techniques and approaches.

How to build and benchmark your LLM evals.

A step-by-step guide to selecting and running your own generative model.

Google Research: Outperforming larger language models with less training data and smaller model sizes - distilling step-by-step.

Google Research: Rethinking calibration for in-context learning and prompt engineering.

Apache Kafka as a mission-critical Data Fabric for GenAI.

Training ChatGPT on your own data.

Hugging Face's guide to optimizing LLMs in production.

Hugging Face is becoming the "GitHub" for Large Language Models.

Building microservice for multi-chat backends using Llama and ChatGPT.

Connect GPT models with company data in Microsoft Azure.

Tuning LLMs with MakerSuite.

Fine-tuning LLMs: Parameter Efficient Fine Tuning (PEFT), LoRA and QLoRA.

How to train BERT for masked language modeling tasks.

Extending context length in Large Language Models.

Conversational applications with Large Language Models understanding the sequence of user inputs, prompts, and responses.

Using data lakes and Large Language Models in development.

How to build an LLM from scratch.

LLM output parsing: function calling vs. LangChain.

Enhancing the power of Llama 2: 3 easy methods for improving your Large Language Model.

Keeping LLMs relevant and current - Retrieval Augmented Generation (RAG).

Build and deploy Retrieval Augmented Generative Pipelines with Haystack.

Why your RAG is not reliable in a production environment.

QCon San Francisco: 

Unlocking enterprise value with Large Language Models.

A modern compute stack for scaling large AI, ML, & LLM workloads.