30 | NoteBookLM: Create an Instant Podcast of your PDFs

Our Menu :)

• Google's NotebookLM: Instant Podcast for YOU

• AI Weekly News Digest

Actionable Tips:

• ChatGPT Prompts for Business

• Create a Synthetic Dataset Using Llama 3.1 to Fine-Tune Your LLM

Jobs and Training

• Anthropic's LLM Bootcamp: Improve Your Prompts

• Job Listings this week

Read time: 4 Minutes

GEN AI AT WORK
Google's NotebookLM: Instant Podcast for YOU 🎧

Buckle up, because Google just dropped a BOMB, that's about to turn your boring old documents into a binge-worthy podcast series. Say hello to NotebookLM, the tool that's putting the 'AI' in 'auditory learning'!

What Does NotebookLM Do?

It uses two AI voices to read your documents, making it feel more conversational. The goal is to make information retention as easy as listening to a podcast.

You read that right! It transforms any document you give it into an INSTANT PODCAST!

How Does It Work?

• AI Model: NotebookLM uses Google's Gemini 1.5, a multimodal AI model, to process the text.

• Text-to-Speech: The AI-powered text-to-speech technology ensures the voices sound natural.

• Summarization: NotebookLM doesn't just read the text; it understands, summarizes, and highlights important information.

• Citations: It includes citations and quotes to back up its content.

Additional Features

• Study Guide: Generates instant summaries for quick review.

• Briefing: Creates quick overviews of documents.

• Privacy: Your personal data is not used for AI training.

Technical Details

• Context window capacity: ~50 PDFs

• Language support: Currently English-only

• Input formats: PDFs, Google Docs, Slides, web URLs

• Output: Downloadable audio file

• NotebookLM's architecture manages large input volumes and cross-references information between multiple documents. 

• However, audio generation for extensive notebooks can take several minutes, indicating computational intensity.

How to

1/ Go to https://notebooklm.google/

2/ On the left panel, click on <Add source>

3/ Drag and drop your document, in my case I have added a scientific paper about <What is the Role of Small Models in the LLM Era: A Survey>, scheduled reading

📝 Paper HERE

4/ Result .. I am dreaming!!!

5/ Chat with your document, the interface is super awesome.

GEN AI PROTECTION
AI Weekly News Digest

[AI Code Agents] Cognition's Devin world's first AI software engineer) significantly improves code generation capability using OpenAI's new 01 model

[Competer Vision] Open-Sourced tool with Github code went viral, that analyze video data of customer behavior in retail space

[OpenAI] OpenAI's o1-preview model ranks 1st officially on Chatbot Arena, across the board, especially in Math, Hard Prompts, and Coding.

ACTIONABLE TIPS
ChatGPT Prompts for Business

1/ PEST Analysis Prompt:

Perform a comprehensive PEST analysis for a [industry] company operating in [location]. Examine the political, economic, social, and technological factors that could impact the business. Provide detailed insights and strategic recommendations based on your analysis."

2/ The Five Whys Prompt:

ChatGPT, use the Five Whys technique to identify the root cause of [Specific Problem] in my business."

3/ Customer Journey Mapping Prompt:

Assist me in mapping the customer journey for a [type of customer] purchasing [product/service]. Identify each touchpoint, the customer's goals and emotions at each stage, and potential pain points. Offer suggestions to enhance the customer experience throughout the journey

4/ Growth Hacking Strategies Prompt:

Suggest innovative growth hacking strategies for a [industry] startup aiming to rapidly increase user acquisition and engagement with a minimal budget. Include actionable tactics and real-world examples of successful implementations

5/ Lean Startup Methodology Prompt:

Explain how to apply the Lean Startup methodology to develop a new product in the [industry] sector. Outline key steps like hypothesis formulation, building a minimum viable product (MVP), testing, and iterating based on customer feedback. Provide tips to maximize efficiency and success during this process.

ACTIONABLE TIPS
Create a Synthetic Dataset Using Llama 3.1 to Fine-Tune Your LLM

This guide will show you how to use Llama 3.1 405B and Nvidia Nemotron 4 to create a synthetic dataset for instruction fine-tuning.

Key Steps

1. Get Access: Obtain an Nvidia NIM API key to access Llama 3.1 405B.

2. Generate Subtopics: Use Llama 3.1 to generate subtopics for your dataset.

3. Create Instructions: Use Llama 3.1 to create instructions for each subtopic.

4. Generate Responses: Use Llama 3.1 to generate responses for each instruction.

5. Score and Filter: Use Nvidia's Nemotron 4 to score and filter the responses.

6. Remove Low-Quality Responses: Remove responses based on helpfulness and verbosity scores.

7. Save Dataset: Save the filtered dataset as a JSON Lines file.

8. Upload Dataset: Log in to Hugging Face and upload your dataset.

This process helps you create a synthetic dataset for fine-tuning smaller LLMs. It uses Llama 3.1 405B to generate high-quality instruction-response pairs and Nemotron 4 to refine the dataset.

This method is cost-effective and gives you control over the dataset creation process.

JOBS AND TRAININGS
Anthropic's LLM Bootcamp: Improve Your Prompts

Anthropic has released a free course on LLM prompt evaluations to enhance your AI applications.

What's the Big Deal?

If your AI isn't performing well, LLM prompt evaluations can help. These evaluations include:

• Systematic tests to assess your prompts

• Performance measurements to improve AI capabilities

• Methods to significantly boost AI performance

Course Overview

The course consists of 9 chapters, covering techniques used by Anthropic with enterprise clients. Topics include:

• Test Dataset Mastery

• Exact String Matching

• Keyword Detective Work

• Regex Wizardry

• LLM Grading Guru

• Custom Rubric Design

• Prompt Iteration Station

• Model Version Showdown

• Quality Gatekeeper

Why Should You Care?

• Improve AI performance

• Free course

JOBS AND TRAININGS

Glean is seeking an Enterprise Account Executive for their team.

Cohere is looking for a Machine Learning Intern/Co-op to join their team.

Meta is hiring a Technical Program Manager for their Reality Labs Research.

Mistral AI is looking for a Machine Learning Engineer to join their team.

Thank you, see you next week!