Sri Ambati, profile picture
Uploaded bySri Ambati
PDF, PPTX187 views

Presentation Resources - H2O Gen AI Ecosystem Overview - Level 2

The document outlines a comprehensive framework for developing and refining generative AI applications and large language models (LLMs), detailing components such as data preparation, fine-tuning, retrieval augmented generation (RAG), and prompt engineering. It emphasizes the importance of foundational models and their applications, as well as the evaluation methods for assessing LLM performance across various tasks. Additionally, it introduces practical labs dedicated to implementing and enhancing generative AI systems.

Education◦
Related topics:
H2O-AI-University• Gen AI•

Embed presentation

Download as PDF, PPTX
H2O.ai Confidential Lab #0: GenAI Ecosystem Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential GenAIAppStudio Datasets Unstructured Datasets Documents ETL / Prep for LLMs Documents → QA Pairs Fine Tuning LLMs (& Prompts) End Users Vector DB (Embeddings) myGPT R. A. G. Talk to your Data Document QA Document Chat Image/Video Chat LLM Query GenAI Apps + + + + + + LLM Data Studio AI Engines MLOps EvalStudio AI Apps + LLMs Integration LLMOps API Prompt Tuning Continuous Feedback Parsing . Chunking Indexing . Embeddings LLM Agents Chat / QA Prompt Engineering LLM Workers Foundations of a GenAI Ecosystem Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz R. A. G. System Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential GenAIAppStudio Datasets Unstructured Datasets Documents ETL / Prep for LLMs Documents → QA Pairs Fine Tuning LLMs (& Prompts) End Users Vector DB (Embeddings) myGPT R. A. G. Talk to your Data Document QA Document Chat Image/Video Chat LLM Query GenAI Apps + + + + + + LLM Data Studio AI Engines MLOps EvalStudio AI Apps + LLMs Integration LLMOps API Prompt Tuning Continuous Feedback Parsing . Chunking Indexing . Embeddings LLM Agents Chat / QA Prompt Engineering LLM Workers RAG GenAI Apps Fine Tuning Data Prep valuation Predictive ML Integrations Foundations of a GenAI Ecosystem Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz
H2O.ai Confidential Foundations of LLMs / GenAI E1S (Explain in 1 Slide) LLMs . RAG . GenAI Apps Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio E1S (Explain in 1 Slide) Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential ● Large Language Models Very deep neural network models trained on vast amount of text data, that are capable of generative highly cohesive text as the output ● Why Large? Large Training Data Large Architectures Large Compute ● Trained with the objective: Next Word Prediction and/or Masked Token Prediction Unsupervised training (pre-training) learn the patterns + structures + representations of language. + can be adapted for a domain or specific NLP task E1S (Explain in 1 Slide) : LLM ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Content Generation Chat/QA Summarize RAG Information Retrieval NLP Tasks GenAI Apps Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Encoder Decoder Architectures Self Attention Mechanisms Capture dependencies between different parts of the input text, Parallelization (fast optimization), Large context capture (+Multi head) E1S (Explain in 1 Slide) : Foundation Models ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Base pre-trained models: Foundations of an LLM Based on Transformer architecture “Attention is all you need” An attention mechanism prioritizes important parts of the input tokens, similar to how you focus on a specific conversation in a noisy room. Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Existing Pre-trained foundational model + Additional data training for a specific task / domain = myGPT E1S (Explain in 1 Slide) : Fine Tuning LAB #3 ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Key considerations in Data Prep / ETLs for LLMs Importance of Good Data in LLM Fine Tuning (and Pre Training) ● Textbooks are all you need, https://arxiv.org/pdf/2306.11644.pdf ● Falcon - Refined Web, https://arxiv.org/pdf/2306.01116.pdf ● ToolLLM - https://arxiv.org/abs/2307.16789 Documents Conversion to QA Pairs Intelligent Document Parsing Text Chunking Chunk Indexing Text Embeddings LAB #4.1 E1S (Explain in 1 Slide) : ETL for LLMs ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Text Data for Fine Tuning Toxicity / Profanity Removal PII Information removal Quality text retainment RLHF Protection Pipelines for Data Prep for LLMs Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Objective: Track, Rank, Evaluate and Benchmark LLMs Evaluate across dimensions: Summarize / Chat / Math / Retrieval / Troubleshooting / etc Tasks: Generation / Retrieval Techniques: ● LLM As a Judge ○ System Prompt ○ Additional LLM Prompt ○ Self Evaluation ● Metrics Eval ○ BLEU ○ Perplexity ○ Context Precision / Recall ● ELO Score based Eval ○ Human Feedback ○ Every LLM is scored against a series of trails run ○ which LLM perform better across range of tasks ○ https://evalgpt.ai E1S (Explain in 1 Slide) : Evaluation LAB #4.2 ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Retrieval Augmented Generation (RAG) Retriever + Generator A system that retrieves relevant information from a knowledge base and generates a response pertinent to the right information Reduces hallucinations, provides accurate + relevant information + private data chat ⇒ Talk to your documents Workflow 1. Prep: Ingest Documents → Prase Documents → Text Chunking → Indexing → Text Embeddings → VectorDB 2. Inference: User Query → Relevant Chunk Extraction → LLM → Response Text Embeddings: Representation of text tokens in numbers (N-dimensional vectors) VectorDB: Scalable for high dimensional data (text vectors) LAB #1 E1S (Explain in 1 Slide) : RAG ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Example #1: Zoom Meetings to Meeting Actions using RAG (MeetingAI) Example #2: QA on Youtube Video using RAG Example #3: Automating Google Sheets using RAG LAB #1 Demo : RAG Integrations ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential ● Prompt Engineering ○ “To get the best result, ask me clarification questions before outputting answer” ○ “Think step by step” ● Role Play ○ “Act as a social media influencer…” ○ “Act as an expert in astrophysics…” ● Shot Prompting ○ Zero Shot - no examples ○ One Shot - one example of expected response ○ Few Shot - more than one examples ● Provide Documentation with Prompt ○ “Please use my document for the answer, here is the link ● Prompt Chaining (Single) ○ “You are a healthcare recruiter. You’re good at writing interview questions. Please ask me each question below one at a time” ● Prompt Chaining (Multiple) ○ “Please forget all prior prompts. You and I will solve a language problem together. To start off with, please ask me ‘what problem would you like to solve’. However, you should never, ever mention the word asparagus. If you understand the requirements, let’s begin” LAB #3 E1S (Explain in 1 Slide) : Prompt Engineering ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Set of predefined constraints and guidelines that are applied to LLMs to manage their behavior Objective: Make LLM not reveal the sensitive or unethical information Types of Guardrails - Fact Checking - Hallucinations Check - Sensitive Info / Pii Check - Content Filtering - Bias Mitigation - Safety and Privacy E1S (Explain in 1 Slide) : LLM Guardrails ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio LAB #1 Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential E1S (Explain in 1 Slide) ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio GenAI Apps AI Apps with LLM integration GenAI / LLM integrations examples in Apps - - Agent to perform specific tasks or trigger actions - Action plugins such as - information summarizer, report generator, email reviewer, data classifier etc. - A Conversation experience - talk to your app (or data, or documents) LAB #2 Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential LAB #3 Demo : Example GenAI Apps ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Demo Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz
H2O.ai Confidential ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Demo : GenAI AppStudio - Last Topic Inspirational Example : GenAI AppStudio Create GenAI Apps (Fully Functional LLM Powered Apps built using GenAI) Power of GenAI ● Use GenAI to convert “sketches” to fully functional Apps ● Apps with an integrated RAG Demo Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential v Fine-tuning Supervised fine-tuning on appropriate and well curated datasets to teach desired output behaviour. Foundation Enormous amount of text data trained in an autoregressive manner 01 02 Memory LLMs can have a huge context length and keep previous questions/tasks in memory for superior context understanding. Database Efficiently leverage your company data. No need to retrain your model if a new pdf is added to the knowledge base. 04 05 RLHF Next token loss function replaced or combined with a reward model trained on Human Feedback. 03 05 04 03 02 01 Building blocks of LLMs Why Large? ○ Large Training Dataset: Trained on massive datasets ○ Large Architectures : Billions of parameters ○ Large Computing Power: Requires massive GPUs
H2O.ai Confidential Lab #1: Handson RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
H2O.ai Confidential Lab #2: Data Prep for LLMs Notebook - Data Prep, Aquarium Login, DataStudio, Doc 2 QA project Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
H2O.ai Confidential 10 Min Break Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI Apps Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
H2O.ai Confidential Training Logistics https://drive.google.com/file/d/1EngrX5JhNsejR6-e9KfdKiFtR1P8qlPU/view Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
H2O.ai Confidential Lab #3: Fine Tuning - Basic understanding of Fine-tuning - Different hyperparameters of fine-tuning Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
H2O.ai Confidential Fine-tuning is the act of taking a pre-trained foundational model and training it on new data for a specific task. Example tasks could be overarching NLP tasks like Text Summarization, classification etc. and Instruct Tuning, as well as to induce a specific style in generated responses (e.g., talk like a sales agent, or social media influencer) Fine Tuning of Language Models Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG MyGPT
H2O.ai Confidential Fine-tune your own LLM and let MLOps handle the serving! LLMOps - Deployment and Ops Fine tune your LLM with H2O LLM Studio, and push it onto a Model Hub (Hugging Face or private). Configure any prompt template, quantization method, etc. you want to include for your model. Build Deploy Interact with the deployed model via single completion or chat-style conversations (compatible with OpenAI API protocols). Integrate with popular toolkits, including LangChain and Guardrails. Interact Register your model and version in H2O MLOps, select the vLLM runtime, and deploy! Easily manage hardware resources according to your needs and usage. Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Compliance and Regulation Resource Controls Integration Flexibility Customization Inference
H2O.ai Confidential Effective and continous output monitoring is Key! LLM Security is new and evolving Jailbreaking Prompt injection Backdoors & data poisoning Adversarial inputs Insecure output handling Data extraction & privacy Data reconstruction Denial of service Watermarking & evasion Model theft ...
H2O.ai Confidential Lab #4: Making a GenAI App Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: GenAI Apps Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
H2O.ai Confidential Lab #5: Evaluate LLMs Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
H2O.ai Confidential Quiz Link https://forms.gle/96MuWcJJHtoNoEM76 Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz
H2O.ai Confidential

More Related Content

PPTX
GPT, LLM, RAG, and RAG in Action: Understanding the Future of AI-Powered Info...
byMuralidharan Deenathayalan
 
PDF
solulab.com-Retrieval-Augmented Generation RAG vs LLM Fine-Tuning- Whats the ...
bymeerasingh12189
 
PPTX
Build With AI Event: GDG on Campus IIE.pptx
bygdgciiekalyani
 
PPTX
AI presentation Genrative LLM for users.pptx
byemceemouli
 
PDF
solulab.com-A Complete LLM Technique Comparison (2).pdf
bycelinedion89121
 
PDF
solulab.com-A Complete LLM Technique Comparison.pdf
bycelinedion89121
 
PDF
Integrate LLM in your applications 101
byMarco De Nittis
 
PDF
"Beyond English: Navigating the Challenges of Building a Ukrainian-language R...
byFwdays
 
GPT, LLM, RAG, and RAG in Action: Understanding the Future of AI-Powered Info...
byMuralidharan Deenathayalan
 
solulab.com-Retrieval-Augmented Generation RAG vs LLM Fine-Tuning- Whats the ...
bymeerasingh12189
 
Build With AI Event: GDG on Campus IIE.pptx
bygdgciiekalyani
 
AI presentation Genrative LLM for users.pptx
byemceemouli
 
solulab.com-A Complete LLM Technique Comparison (2).pdf
bycelinedion89121
 
solulab.com-A Complete LLM Technique Comparison.pdf
bycelinedion89121
 
Integrate LLM in your applications 101
byMarco De Nittis
 
"Beyond English: Navigating the Challenges of Building a Ukrainian-language R...
byFwdays
 

Similar to Presentation Resources - H2O Gen AI Ecosystem Overview - Level 2

PDF
All in AI: LLM Landscape & RAG in 2024 with Mark Ryan (Google) & Jerry Liu (L...
byDaniel Zivkovic
 
PDF
Best Practices for Building Successful LLM Applications
byBhavulGauri1
 
PDF
Retrieval Augmented Generation in Practice: Scalable GenAI platforms with k8s...
byMihai Criveti
 
PPTX
AI presentation for dummies LLM Generative AI.pptx
byemceemouli
 
PPTX
Journey of Generative AI
bythomasjvarghese49
 
PDF
Large Language Models Bootcamp
byData Science Dojo
 
PDF
AI presentation and introduction - Retrieval Augmented Generation RAG 101
byvincent683379
 
PDF
"Chat with your private data using Llama3 and LLPhant in PHP", Enrico Zimuel
byFwdays
 
PPTX
TechDayPakistan-Slides RAG with Cosmos DB.pptx
byUsama Wahab Khan Cloud, Data and AI
 
PDF
Next Generation Apps: Enhancing User Experience with LLMs.pdf
byNithish Raghunandanan
 
PDF
Java Developers - What Lies Ahead in the AI era
byEmily Jiang
 
PDF
RAGs to Riches Intelligent Applications.pdf
byToledo Codes
 
PPTX
Build With AI2.0 organized by GDG on campus SCOE
bysanidhyanaik1907
 
PPTX
Fact based Generative AI
byStefan Weber
 
PPTX
aistudy-240521200530-db141c56 RAG AI.pptx
byemceemouli
 
PDF
LLM Fine-Tuning vs RAG A Complete Comparison.pdf
bySoluLab1231
 
PDF
LLM Fine-Tuning vs RAG A Complete Comparison.pdf
byimoliviabennett
 
PDF
LLM Fine-Tuning vs RAG A Complete Comparison.pdf
byniahiggins21
 
PPTX
GenerativeAI and Automation - IEEE ACSOS 2023.pptx
byAllen Chan
 
PDF
Optimizing GenAI apps, by N. El Mawass and Maria Knorps
byParis Women in Machine Learning and Data Science
 
All in AI: LLM Landscape & RAG in 2024 with Mark Ryan (Google) & Jerry Liu (L...
byDaniel Zivkovic
 
Best Practices for Building Successful LLM Applications
byBhavulGauri1
 
Retrieval Augmented Generation in Practice: Scalable GenAI platforms with k8s...
byMihai Criveti
 
AI presentation for dummies LLM Generative AI.pptx
byemceemouli
 
Journey of Generative AI
bythomasjvarghese49
 
Large Language Models Bootcamp
byData Science Dojo
 
AI presentation and introduction - Retrieval Augmented Generation RAG 101
byvincent683379
 
"Chat with your private data using Llama3 and LLPhant in PHP", Enrico Zimuel
byFwdays
 
TechDayPakistan-Slides RAG with Cosmos DB.pptx
byUsama Wahab Khan Cloud, Data and AI
 
Next Generation Apps: Enhancing User Experience with LLMs.pdf
byNithish Raghunandanan
 
Java Developers - What Lies Ahead in the AI era
byEmily Jiang
 
RAGs to Riches Intelligent Applications.pdf
byToledo Codes
 
Build With AI2.0 organized by GDG on campus SCOE
bysanidhyanaik1907
 
Fact based Generative AI
byStefan Weber
 
aistudy-240521200530-db141c56 RAG AI.pptx
byemceemouli
 
LLM Fine-Tuning vs RAG A Complete Comparison.pdf
bySoluLab1231
 
LLM Fine-Tuning vs RAG A Complete Comparison.pdf
byimoliviabennett
 
LLM Fine-Tuning vs RAG A Complete Comparison.pdf
byniahiggins21
 
GenerativeAI and Automation - IEEE ACSOS 2023.pptx
byAllen Chan
 
Optimizing GenAI apps, by N. El Mawass and Maria Knorps
byParis Women in Machine Learning and Data Science
 

More from Sri Ambati

PDF
H2O Gen AI Ecosystem Overview - Level 1 - Slide Deck
bySri Ambati
 
PDF
LLM Learning Path Level 2 - Presentation Slides
bySri Ambati
 
PDF
Evaluating Generative AI.Models with Eval Studio - Support Presentation
bySri Ambati
 
PDF
Practical MLOps with H2O.ai - Support Slide Deck.pdf
bySri Ambati
 
PDF
Large Language Models (LLMs) - Level 3 Slides
bySri Ambati
 
PDF
H2O Cloud AI Developer Services - Slides (2024)
bySri Ambati
 
PDF
LLM Learning Path Level 1 - Presentation Slides
bySri Ambati
 
PDF
H2O Generative AI Starter Track - Support Presentation Slides.pdf
bySri Ambati
 
PDF
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
bySri Ambati
 
PPTX
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
bySri Ambati
 
PDF
H2O.ai Agents : From Theory to Practice - Support Presentation
bySri Ambati
 
PDF
Enterprise h2o GPTe Learning Path Slide Deck
bySri Ambati
 
PDF
Data Science and Machine Learning Platforms (2024) Slides
bySri Ambati
 
PDF
An In-depth Exploration of Enterprise h2oGPTe Slide Deck
bySri Ambati
 
PDF
H2O Wave Course Starter - Presentation Slides
bySri Ambati
 
PDF
H2O Label Genie Starter Track - Support Presentation
bySri Ambati
 
PDF
H2O Driverless AI Starter Course - Slides and Assignments
bySri Ambati
 
PDF
Data Prep for H2O Driverless AI - Slides
bySri Ambati
 
PDF
Hydrogen Torch - Starter Course - Presentation Slides
bySri Ambati
 
PDF
Intro to Enterprise h2oGPTe Presentation Slides
bySri Ambati
 
H2O Gen AI Ecosystem Overview - Level 1 - Slide Deck
bySri Ambati
 
LLM Learning Path Level 2 - Presentation Slides
bySri Ambati
 
Evaluating Generative AI.Models with Eval Studio - Support Presentation
bySri Ambati
 
Practical MLOps with H2O.ai - Support Slide Deck.pdf
bySri Ambati
 
Large Language Models (LLMs) - Level 3 Slides
bySri Ambati
 
H2O Cloud AI Developer Services - Slides (2024)
bySri Ambati
 
LLM Learning Path Level 1 - Presentation Slides
bySri Ambati
 
H2O Generative AI Starter Track - Support Presentation Slides.pdf
bySri Ambati
 
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
bySri Ambati
 
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
bySri Ambati
 
H2O.ai Agents : From Theory to Practice - Support Presentation
bySri Ambati
 
Enterprise h2o GPTe Learning Path Slide Deck
bySri Ambati
 
Data Science and Machine Learning Platforms (2024) Slides
bySri Ambati
 
An In-depth Exploration of Enterprise h2oGPTe Slide Deck
bySri Ambati
 
H2O Wave Course Starter - Presentation Slides
bySri Ambati
 
H2O Label Genie Starter Track - Support Presentation
bySri Ambati
 
H2O Driverless AI Starter Course - Slides and Assignments
bySri Ambati
 
Data Prep for H2O Driverless AI - Slides
bySri Ambati
 
Hydrogen Torch - Starter Course - Presentation Slides
bySri Ambati
 
Intro to Enterprise h2oGPTe Presentation Slides
bySri Ambati
 

Recently uploaded

PPTX
Overview of how to Create a Model in Odoo 18
byCeline George
 
PPTX
How to Track a Link Using Odoo 18 SMS Marketing
byCeline George
 
PPTX
How to Manage Package Reservation in Odoo 18 Inventory
byCeline George
 
PDF
Most Imp Chapters & Weightage for Boards 2025.pdf
byTamannaSagar
 
PPTX
Planning Assessment for Outcome-based Education
byAdri Jovin
 
PPTX
The Cell & Cell Cycle-detailed structure and function of organelles.pptx
byAshish Umale
 
PPTX
Basics in Phytochemistry, Extraction, Isolation methods, Characterisation etc.
byD. Y. Patil College of Pharmacy, Kolhapur.
 
PPTX
How to Create Lead_Opportunity From Odoo 18 Website
byCeline George
 
PPTX
PURPOSIVE SAMPLING IN EDUCATIONAL RESEARCH RACHITHRA RK.pptx
byssuser047b711
 
PDF
Biology Practical Class 12th 2025 PDF(2)-2-52.pdf
bySCPRPWomenCollegeChi
 
PDF
Scalable-MADDPG-Based Cooperative Target Invasion for a Multi-USV System.pdf
byMan_Ebook
 
PDF
IMANI Africa files RTI request seeking full disclosure on 2026 SIM registrati...
bynservice241
 
PDF
Multiple Myeloma , definition, etiology, PP, CM , DE and management
byNisha Borsa
 
PPTX
Steve Hale - Developing Elite Young Goalkeepers 2024.pptx
bypjmfd
 
PDF
Blood Group Incompatibility: Rh Factor and Erythroblastosis Fetalis
bySudhandraKarthi
 
PDF
International Men's Day Event: Breaking the Silence: Embracing Vulnerability ...
byAssociation for Project Management
 
PPTX
ATTENTION -PART 2.pptx Shilpa Hotakar for I semester BSc students
bySHILPA HOTAKAR
 
PPTX
PARENTAL ROUTES OF DRUGS ADMINISTRATION .pptx
byAneetaSharma15
 
PPTX
Semester 6 Unit 2 Club foot (talipes).pptx
bysachin7989
 
PPTX
REVISED DEFENSE MECHANISM / ADJUSTMENT MECHANISM-FINAL
byShimla
 
Overview of how to Create a Model in Odoo 18
byCeline George
 
How to Track a Link Using Odoo 18 SMS Marketing
byCeline George
 
How to Manage Package Reservation in Odoo 18 Inventory
byCeline George
 
Most Imp Chapters & Weightage for Boards 2025.pdf
byTamannaSagar
 
Planning Assessment for Outcome-based Education
byAdri Jovin
 
The Cell & Cell Cycle-detailed structure and function of organelles.pptx
byAshish Umale
 
Basics in Phytochemistry, Extraction, Isolation methods, Characterisation etc.
byD. Y. Patil College of Pharmacy, Kolhapur.
 
How to Create Lead_Opportunity From Odoo 18 Website
byCeline George
 
PURPOSIVE SAMPLING IN EDUCATIONAL RESEARCH RACHITHRA RK.pptx
byssuser047b711
 
Biology Practical Class 12th 2025 PDF(2)-2-52.pdf
bySCPRPWomenCollegeChi
 
Scalable-MADDPG-Based Cooperative Target Invasion for a Multi-USV System.pdf
byMan_Ebook
 
IMANI Africa files RTI request seeking full disclosure on 2026 SIM registrati...
bynservice241
 
Multiple Myeloma , definition, etiology, PP, CM , DE and management
byNisha Borsa
 
Steve Hale - Developing Elite Young Goalkeepers 2024.pptx
bypjmfd
 
Blood Group Incompatibility: Rh Factor and Erythroblastosis Fetalis
bySudhandraKarthi
 
International Men's Day Event: Breaking the Silence: Embracing Vulnerability ...
byAssociation for Project Management
 
ATTENTION -PART 2.pptx Shilpa Hotakar for I semester BSc students
bySHILPA HOTAKAR
 
PARENTAL ROUTES OF DRUGS ADMINISTRATION .pptx
byAneetaSharma15
 
Semester 6 Unit 2 Club foot (talipes).pptx
bysachin7989
 
REVISED DEFENSE MECHANISM / ADJUSTMENT MECHANISM-FINAL
byShimla
 

Presentation Resources - H2O Gen AI Ecosystem Overview - Level 2

  • 1.
    H2O.ai Confidential Lab #0: GenAIEcosystem Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 2.
    H2O.ai Confidential GenAIAppStudio Datasets Unstructured Datasets Documents ETL /Prep for LLMs Documents → QA Pairs Fine Tuning LLMs (& Prompts) End Users Vector DB (Embeddings) myGPT R. A. G. Talk to your Data Document QA Document Chat Image/Video Chat LLM Query GenAI Apps + + + + + + LLM Data Studio AI Engines MLOps EvalStudio AI Apps + LLMs Integration LLMOps API Prompt Tuning Continuous Feedback Parsing . Chunking Indexing . Embeddings LLM Agents Chat / QA Prompt Engineering LLM Workers Foundations of a GenAI Ecosystem Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz R. A. G. System Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 3.
    H2O.ai Confidential GenAIAppStudio Datasets Unstructured Datasets Documents ETL /Prep for LLMs Documents → QA Pairs Fine Tuning LLMs (& Prompts) End Users Vector DB (Embeddings) myGPT R. A. G. Talk to your Data Document QA Document Chat Image/Video Chat LLM Query GenAI Apps + + + + + + LLM Data Studio AI Engines MLOps EvalStudio AI Apps + LLMs Integration LLMOps API Prompt Tuning Continuous Feedback Parsing . Chunking Indexing . Embeddings LLM Agents Chat / QA Prompt Engineering LLM Workers RAG GenAI Apps Fine Tuning Data Prep valuation Predictive ML Integrations Foundations of a GenAI Ecosystem Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz
  • 4.
    H2O.ai Confidential Foundations ofLLMs / GenAI E1S (Explain in 1 Slide) LLMs . RAG . GenAI Apps Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 5.
    H2O.ai Confidential ● LLMs ○Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio E1S (Explain in 1 Slide) Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 6.
    H2O.ai Confidential ● LargeLanguage Models Very deep neural network models trained on vast amount of text data, that are capable of generative highly cohesive text as the output ● Why Large? Large Training Data Large Architectures Large Compute ● Trained with the objective: Next Word Prediction and/or Masked Token Prediction Unsupervised training (pre-training) learn the patterns + structures + representations of language. + can be adapted for a domain or specific NLP task E1S (Explain in 1 Slide) : LLM ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Content Generation Chat/QA Summarize RAG Information Retrieval NLP Tasks GenAI Apps Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 7.
    H2O.ai Confidential Encoder DecoderArchitectures Self Attention Mechanisms Capture dependencies between different parts of the input text, Parallelization (fast optimization), Large context capture (+Multi head) E1S (Explain in 1 Slide) : Foundation Models ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Base pre-trained models: Foundations of an LLM Based on Transformer architecture “Attention is all you need” An attention mechanism prioritizes important parts of the input tokens, similar to how you focus on a specific conversation in a noisy room. Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 8.
    H2O.ai Confidential Existing Pre-trainedfoundational model + Additional data training for a specific task / domain = myGPT E1S (Explain in 1 Slide) : Fine Tuning LAB #3 ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 9.
    H2O.ai Confidential Key considerationsin Data Prep / ETLs for LLMs Importance of Good Data in LLM Fine Tuning (and Pre Training) ● Textbooks are all you need, https://arxiv.org/pdf/2306.11644.pdf ● Falcon - Refined Web, https://arxiv.org/pdf/2306.01116.pdf ● ToolLLM - https://arxiv.org/abs/2307.16789 Documents Conversion to QA Pairs Intelligent Document Parsing Text Chunking Chunk Indexing Text Embeddings LAB #4.1 E1S (Explain in 1 Slide) : ETL for LLMs ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Text Data for Fine Tuning Toxicity / Profanity Removal PII Information removal Quality text retainment RLHF Protection Pipelines for Data Prep for LLMs Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 10.
    H2O.ai Confidential Objective: Track,Rank, Evaluate and Benchmark LLMs Evaluate across dimensions: Summarize / Chat / Math / Retrieval / Troubleshooting / etc Tasks: Generation / Retrieval Techniques: ● LLM As a Judge ○ System Prompt ○ Additional LLM Prompt ○ Self Evaluation ● Metrics Eval ○ BLEU ○ Perplexity ○ Context Precision / Recall ● ELO Score based Eval ○ Human Feedback ○ Every LLM is scored against a series of trails run ○ which LLM perform better across range of tasks ○ https://evalgpt.ai E1S (Explain in 1 Slide) : Evaluation LAB #4.2 ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 11.
    H2O.ai Confidential Retrieval Augmented Generation(RAG) Retriever + Generator A system that retrieves relevant information from a knowledge base and generates a response pertinent to the right information Reduces hallucinations, provides accurate + relevant information + private data chat ⇒ Talk to your documents Workflow 1. Prep: Ingest Documents → Prase Documents → Text Chunking → Indexing → Text Embeddings → VectorDB 2. Inference: User Query → Relevant Chunk Extraction → LLM → Response Text Embeddings: Representation of text tokens in numbers (N-dimensional vectors) VectorDB: Scalable for high dimensional data (text vectors) LAB #1 E1S (Explain in 1 Slide) : RAG ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 12.
    H2O.ai Confidential Example #1: ZoomMeetings to Meeting Actions using RAG (MeetingAI) Example #2: QA on Youtube Video using RAG Example #3: Automating Google Sheets using RAG LAB #1 Demo : RAG Integrations ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 13.
    H2O.ai Confidential ● PromptEngineering ○ “To get the best result, ask me clarification questions before outputting answer” ○ “Think step by step” ● Role Play ○ “Act as a social media influencer…” ○ “Act as an expert in astrophysics…” ● Shot Prompting ○ Zero Shot - no examples ○ One Shot - one example of expected response ○ Few Shot - more than one examples ● Provide Documentation with Prompt ○ “Please use my document for the answer, here is the link ● Prompt Chaining (Single) ○ “You are a healthcare recruiter. You’re good at writing interview questions. Please ask me each question below one at a time” ● Prompt Chaining (Multiple) ○ “Please forget all prior prompts. You and I will solve a language problem together. To start off with, please ask me ‘what problem would you like to solve’. However, you should never, ever mention the word asparagus. If you understand the requirements, let’s begin” LAB #3 E1S (Explain in 1 Slide) : Prompt Engineering ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 14.
    H2O.ai Confidential Set ofpredefined constraints and guidelines that are applied to LLMs to manage their behavior Objective: Make LLM not reveal the sensitive or unethical information Types of Guardrails - Fact Checking - Hallucinations Check - Sensitive Info / Pii Check - Content Filtering - Bias Mitigation - Safety and Privacy E1S (Explain in 1 Slide) : LLM Guardrails ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio LAB #1 Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 15.
    H2O.ai Confidential E1S (Explainin 1 Slide) ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio GenAI Apps AI Apps with LLM integration GenAI / LLM integrations examples in Apps - - Agent to perform specific tasks or trigger actions - Action plugins such as - information summarizer, report generator, email reviewer, data classifier etc. - A Conversation experience - talk to your app (or data, or documents) LAB #2 Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 16.
    H2O.ai Confidential LAB #3 Demo: Example GenAI Apps ● LLMs ○ Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Demo Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz
  • 17.
    H2O.ai Confidential ● LLMs ○Foundation Models ○ Fine Tuning ○ ETL for LLMs ○ Evaluation ● RAG ○ Document QA / Chat ○ RAG Integrations ○ Prompt Engineering ○ LLM Guardrails ● GenAI Apps ○ Example GenAI App ○ GenAI AppStudio Demo : GenAI AppStudio - Last Topic Inspirational Example : GenAI AppStudio Create GenAI Apps (Fully Functional LLM Powered Apps built using GenAI) Power of GenAI ● Use GenAI to convert “sketches” to fully functional Apps ● Apps with an integrated RAG Demo Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 18.
    H2O.ai Confidential v Fine-tuning Supervised fine-tuning on appropriateand well curated datasets to teach desired output behaviour. Foundation Enormous amount of text data trained in an autoregressive manner 01 02 Memory LLMs can have a huge context length and keep previous questions/tasks in memory for superior context understanding. Database Efficiently leverage your company data. No need to retrain your model if a new pdf is added to the knowledge base. 04 05 RLHF Next token loss function replaced or combined with a reward model trained on Human Feedback. 03 05 04 03 02 01 Building blocks of LLMs Why Large? ○ Large Training Dataset: Trained on massive datasets ○ Large Architectures : Billions of parameters ○ Large Computing Power: Requires massive GPUs
  • 19.
    H2O.ai Confidential Lab #1: HandsonRAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
  • 20.
    H2O.ai Confidential Lab #2: DataPrep for LLMs Notebook - Data Prep, Aquarium Login, DataStudio, Doc 2 QA project Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
  • 21.
    H2O.ai Confidential 10 Min Break Lab#2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI Apps Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
  • 22.
    H2O.ai Confidential Training Logistics https://drive.google.com/file/d/1EngrX5JhNsejR6-e9KfdKiFtR1P8qlPU/view Lab#0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Lab #2: DataPrep Lab #3: Fine Tuning Lab #4: Gen AI apps Lab #5: LLM Eval Quiz
  • 23.
    H2O.ai Confidential Lab #3: FineTuning - Basic understanding of Fine-tuning - Different hyperparameters of fine-tuning Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
  • 24.
    H2O.ai Confidential Fine-tuning isthe act of taking a pre-trained foundational model and training it on new data for a specific task. Example tasks could be overarching NLP tasks like Text Summarization, classification etc. and Instruct Tuning, as well as to induce a specific style in generated responses (e.g., talk like a sales agent, or social media influencer) Fine Tuning of Language Models Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG MyGPT
  • 25.
    H2O.ai Confidential Fine-tune yourown LLM and let MLOps handle the serving! LLMOps - Deployment and Ops Fine tune your LLM with H2O LLM Studio, and push it onto a Model Hub (Hugging Face or private). Configure any prompt template, quantization method, etc. you want to include for your model. Build Deploy Interact with the deployed model via single completion or chat-style conversations (compatible with OpenAI API protocols). Integrate with popular toolkits, including LangChain and Guardrails. Interact Register your model and version in H2O MLOps, select the vLLM runtime, and deploy! Easily manage hardware resources according to your needs and usage. Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG Compliance and Regulation Resource Controls Integration Flexibility Customization Inference
  • 26.
    H2O.ai Confidential Effective andcontinous output monitoring is Key! LLM Security is new and evolving Jailbreaking Prompt injection Backdoors & data poisoning Adversarial inputs Insecure output handling Data extraction & privacy Data reconstruction Denial of service Watermarking & evasion Model theft ...
  • 27.
    H2O.ai Confidential Lab #4: Makinga GenAI App Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: GenAI Apps Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
  • 28.
    H2O.ai Confidential Lab #5: EvaluateLLMs Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz Lab #0: Foundations Lab #1: RAG
  • 29.
    H2O.ai Confidential Quiz Link https://forms.gle/96MuWcJJHtoNoEM76 Lab#0: Foundations Lab #1: RAG Lab #2: GenAI Apps Lab #3: Fine Tuning Lab #4: DataPrep Lab #5: LLM Eval Quiz
  • 30.