Large Language Models - Level 2

Name: Large Language Models - Level 2
Rating: 4.3 (73 reviews)

Master Data Prep, Fine-Tuning for Advanced NLP, and more!

Created byH2O.ai University, Andreea Turcu

Last updated 7/2024

English

What you'll learn

Master Data Quality for NLP Models
Become an LLM DataStudio Pro
Craft Q&A Datasets
Fine-Tune LLMs for Specialized Tasks

Course content

4 sections • 13 lectures • 3h 9m total length

Mastering Data Prep for Enhanced Language Model Performance9:28
Master data preparation and cleaning to boost LLM performance, reduce bias, and enable reliable, ethical downstream NLP tasks, using LLM DataStudio’s workflows for project management.
Key Functions in Language Model Data Preparation9:20
Master key data preparation functions to maximize language model performance. Explore data object, data augmentation, text cleaning, profanity check, quality checks, length control, and sequence handling for diverse tasks.
LLM DataStudio: Streamline Language Model Data Prep7:08
Streamline data preparation for large language models with LLM DataStudio's no-code tools, including text cleaning, Q&A generation, and quality checks, integrated within the H2O.ai ecosystem for fine-tuning and summaries.

LLM DataStudio Interface & Curation Automation9:24
Explore no-code data curation in LLM DataStudio, turning PDFs, audio, and more into QA datasets via intelligent chunking, embeddings, and prompt engineering with H2OGPT.
Mastering LLM Data Preparation with LLM DataStudio: A Step-by-Step Guide52:46
Explore how to prepare high-quality data for large language models with LLM Data Studio, using a no-code interface to curate, augment, and generate Q&A and summaries.
LLM DataStudio: Projects & Workflow Mastery11:31
Explore LLM DataStudio's projects hub to manage data preparation workflows from intake and assessment to result generation. Build configurable, drag-and-drop workflows for question answering datasets, with JSON and CSV outputs.
LLM DataStudio: Prepping Q&A Datasets7:45
Prepare a context-question-answer dataset in LLM DataStudio by configuring a workflow with augmentation, text cleaning, and quality checks, then run the pipeline and review the csv output.
Fine-Tuning Principles for Large Language Models12:41
Explore fine-tuning principles for large language models, covering data, backbones, quantisation and LoRA, plus hands-on use of LLM Studio and deploying to HuggingFace.
Synthetic Datasets and Language Model Backbones13:35
Explore how synthetic datasets simulate real data, enable controlled experiments and privacy-preserving testing, and how backbones support efficient fine-tuning of large language models.

Fine-Tuning with Quantization and LoRA8:54
Fine-tuning adapts large language models to task-specific data, like dialog data in question-answer pairs. Quantization and LoRA reduce size and compute during fine-tuning, balancing efficiency and accuracy for deployment.
LLM Optimization - Techniques and Insights3:35
Explore quantization, LoRA, pruning, and knowledge distillation to optimize LLMs with architecture adjustments for efficiency. Use benchmarking, iterative retraining, and H2O LLM Studio for RL fine-tuning and model export.
A Journey through H2O.ai's LLM Studio37:42
Explore large language models with LLM Studio, a no-code fine-tuning tool that trains on instruction-output datasets, monitors experiments, compares results, and deploys to Hugging Face.
Deploying LLM Models with H2O LLM Studio5:48
Deploy your fine-tuned model with H2O LLM Studio, export to Hugging Face using a write-enabled API key, and follow steps from viewing experiments to pushing checkpoints and exporting.

Requirements

Basic data science concepts.
No programming experience is needed. You will learn everything you need to know.

Description

Continue your exploration of Large Language Models (LLMs) with Andreea Turcu's foundational Level 2 course! Specially designed for those with foundational knowledge, this course delves deep into optimizing Natural Language Processing (NLP) models through robust data practices.

Discover the critical role of clean data and effective data preparation techniques essential for NLP model quality. Using LLM DataStudio, navigate supported workflows, customize interfaces, and implement quality control measures. Learn to set up projects and leverage collaboration features to enhance team efficiency.

Master QnA dataset creation, ensuring accuracy through validation and quality assurance processes.

Perfect fine-tuning with H2O LLM Studio, where you'll tailor models to specific tasks. Explore workflows, employ data augmentation strategies, and select optimal architectures from pre-trained models.

Delve deeper into advanced techniques like Quantisation and LoRA for model compression, optimizing your NLP applications for real-world deployment.

Earn your LLM Certification Level 2, showcasing your expertise in data preparation, fine-tuning, and model optimization. This certification is ideal for professionals that are aiming to excel in specialized roles within NLP, machine learning, and data engineering.

Join Andreea Turcu in this course and elevate your skills in harnessing LLMs for cutting-edge NLP projects, where you’ll dive into practical applications of language models and supercharge your AI career!

Who this course is for:

NLP Enthusiasts
Data Professionals
Machine Learning Beginners
Business Professionals

Large Language Models - Level 2

What you'll learn

Explore related topics

Course content

Getting Started with LLM Data Prep3 lectures • 26min

Mastering LLM DataStudio6 lectures • 1hr 48min

Fine-Tuning Your Large Language Models4 lectures • 56min

Course Completion Quiz0

Requirements

Description

Who this course is for: