Machine Learning Made Simple

Machine Learning Made Simple

By Saugata ChatterjeeTechnology
Download on the App Store

Machine Learning Made Simple episodes

  • Ep 24: The Future of Speech Recognition: How Transformer is Transforming the Game

    Summary:

    Exploring ASR Technologies: Dive into the world of Automatic Speech Recognition (ASR) and its evolving landscape.

    Comparative Analysis: Unpack the differences and advantages between Word2Vec and OpenAI's Whisper models in speech processing.

    Technical Insights: Understand the role of waveform encoding and Hidden Markov Models (HMMs) in enhancing speech recognition systems.

    Tune in to discover how these cutting-edge technologies are revolutionizing speech recognition for professionals and industry experts in AI. Don’t miss out on the latest advancements—subscribe now!


    AI News:

    1. Introducing Meta Llama 3: The most capable openly available LLM to date
    2. https://www.microsoft.com/en-us/research/project/vasa-1/
    3. AI System Can Detect Parkinson's Disease from Brain Waves, Study Finds
    4. [2404.10981] A Survey on Retrieval-Augmented Text Generation for Large Language Models
    5. [2404.11584] The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Survey

      6. References for main topic:

      1. https://arxiv.org/abs/2006.11477v3
      2. Robust Speech Recognition via Large-Scale Weak Supervision


        3. 26 min
      3. Ep 23: Edge Computing - LLMs in Your Palm

        Episode 22: Edge Computing - LLMs in Your Palm

        AI News:

        1. AI computing firm Lambda raises $320 million in fresh funding
        2. Introducing Stable Audio 2.0 — Stability AI
        3. OpenAI's CTO said she wasn't sure if Sora was trained on YouTube videos. YouTube's CEO says that would be a problem.
        4. 2404.03543 CodeEditorBench: Evaluating Code Editing Capability of Large Language Models
        5. 2404.02806 The RealHumanEval: Evaluating Large Language Models' Abilities to Support Programmers
        6. 2404.02905v1 Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction

          7. References for main topic:

          2404.01744 Octopus v2: On-device language model for super agent

          1. Cloud GPUs:
          2. Cloud GPU Comparison
          3.  The 20 Best Cloud GPU Service Providers [RANKED]


            4. 35 min
          4. Ep 22: How small LLMs are outperforming GPT3 using a Mixture of Experts

            Episode22: How small LLMs (47B) are outperforming GPT3 (185B) using a Mixture of Experts (MoE)

            AI News:

            1. 2402.05120 More Agents Is All You Need

            2. 2403.16971 AIOS: LLM Agent Operating System

            3. 2404.02258 Mixture-of-Depths

            4. Devika GitHub Repository - Devika: An Agentic AI Software Engineer

            5. T-Rex GitHub Repository - T-Rex: A Large-Scale Relation Extraction Framework

            6. WSJ Article on Cognition Labs - A Peter Thiel-backed AI startup, Cognition Labs, seeks $2 billion in valuation

              7. References for main topic:

              1. 1701.06538 Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts

              2. 2006.16668 GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding

              3. 2101.03961 Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity

              4. 2112.06905 GLaM: Efficient Scaling of Language Models with Mixture-of-Experts

              5. 2202.08906 ST-MoE: Designing Stable and Transferable Sparse Expert Models

              6. 2211.15841 MegaBlocks: Efficient Sparse Training with Mixture-of-Experts

              7. 2401.04088 Mixtral of Experts

              8. 1511.07543 Convergent Learning: Do different neural networks learn the same representations?


                9. 52 min
              9. Ep 21: LLM Model Merging

                AI News:

                1. Databricks announces the release of a new LLM (Large Language Model) named DBRX, designed to improve copilot features in DBX.

                2. The research paper "LLM4Decompile: Decompiling Binary Code with Large Language Models" [2403.05286] is released, claiming high-level accuracy in decompiling machine code to high-level language.

                3. GitHub introduces an AI tool for detecting security vulnerabilities within code.

                4. Stability AI experiences a CEO resignation.


                Main Topic: Merging LLM Models

                1. SLERP (Smoothly Large Embeddings for Representation Pooling)

                2. TIES (Technically Intuitive Embedding Space)

                3. Frankenmerges (A new approach to merging models)


                References:

                1. [2403.05286] LLM4Decompile: Decompiling Binary Code with Large Language Models

                2. [2311.03099] Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch

                3. [2306.11644] Textbooks Are All You Need

                4. [1909.11299] Mixout: Effective Regularization to Finetune Large-scale Pretrained Language Models

                5. [1511.07543] Convergent Learning: Do different neural networks learn the same representations?

                6. An empirical analysis of compute-optimal large language model training - Google DeepMind

                  7. 33 min
                7. Machine Learning Made Simple Podcasts - Episode 20

                  Topic: Unlocking Secrets of AI: The Power of GLUE in LLM Evaluation - Part 1


                  Summary:

                  • Dive into the intricacies of GLUE, starting with its foundational sub-tasks.
                  • Explore CoLA and SST-2 tasks, unveiling their significance in AI benchmarking.
                  • Unravel the complexities of similarity and paraphrase tasks, including MRPC, QQP, and STS-B.

                    • Join us as we dissect the pivotal components of the GLUE benchmark, providing a comprehensive understanding of LLM evaluation. Don’t miss this insightful journey into the core of AI's evaluation metrics.


                    If you enjoy this podcast, please:

                    • Subscribe to get notifications for new episodes.
                    • ⁠⁠⁠⁠Follow me on LinkedIn⁠⁠⁠⁠ for the latest in AI/ML papers and discussions.

                      • About:

                      Dive into the magic of machine learning with our podcast, where we unravel the mysteries in a language everyone can groove to! Ideal for the movers and shakers in the tech world – from top-tier execs shaping ML strategies to tech leads leading squads of MLEs. Whether you're an IT pro on the brink of a ML adventure or just someone itching to ride the ML wave, we've got your backstage pass to the world of ML hype! Tune in, turn up, and let's demystify machine learning together! 🚀✨ #MLGroove #DecodeTheHype 🎙️

                      ⁠⁠⁠Legal Disclaimer for Machine Learning Made Simple⁠⁠⁠


                      31 min
                    • Machine Learning Made Simple Podcasts - Episode 19

                      Topic: In-Depth Analysis: GPT-4 and Claude Features Compared

                      Summary:

                      • Context Length Difference: Explore how Claude's larger context length provides a significant advantage over GPT-4.
                      • Test Comparisons: A comprehensive comparison of GPT-4 and Claude on standardized tests like GRE, including reading, writing, math, and coding.
                      • AI Capabilities: Delve into the specific features and capabilities that set GPT-4 and Claude apart in the AI landscape.

                        • Join us as we uncover the intricate details of GPT-4 and Claude, offering a clear perspective for AI professionals and industry experts. Discover the strengths and weaknesses of each model and how they shape the future of AI technology.

                        If you enjoy this podcast, please:

                        • Subscribe to get notifications for new episodes.
                        • ⁠⁠⁠Follow me on LinkedIn⁠⁠⁠ for the latest in AI/ML papers and discussions.

                          • About:

                          Dive into the magic of machine learning with our podcast, where we unravel the mysteries in a language everyone can groove to! Ideal for the movers and shakers in the tech world – from top-tier execs shaping ML strategies to tech leads leading squads of MLEs. Whether you're an IT pro on the brink of a ML adventure or just someone itching to ride the ML wave, we've got your backstage pass to the world of ML hype! Tune in, turn up, and let's demystify machine learning together! 🚀✨ #MLGroove #DecodeTheHype 🎙️

                          ⁠⁠Legal Disclaimer for Machine Learning Made Simple⁠⁠


                          19 min
                        • Machine Learning Made Simple Podcasts - Episode 18

                          Topic: Next-Gen AI: How QLoRa's Quantization Transforms LLM Fine-Tuning

                          Summary:

                          • Quantization Breakthrough: Explore how quantization is revolutionizing AI, making models more efficient.
                          • Memory Mastery: Discover the technique that slashes memory usage by 16x, converting 64-bit floats to 4-bit integers.
                          • Augment, Not Replace: Learn how fine-tuned LLMs are set to augment human capabilities, not replace them.

                            • If you enjoy this podcast, please:

                            • Subscribe to get notifications for new episodes.
                            • ⁠⁠Follow me on LinkedIn⁠⁠ for the latest in AI/ML papers and discussions.

                              • About:

                              Dive into the magic of machine learning with our podcast, where we unravel the mysteries in a language everyone can groove to! Ideal for the movers and shakers in the tech world – from top-tier execs shaping ML strategies to tech leads leading squads of MLEs. Whether you're an IT pro on the brink of a ML adventure or just someone itching to ride the ML wave, we've got your backstage pass to the world of ML hype! Tune in, turn up, and let's demystify machine learning together! 🚀✨ #MLGroove #DecodeTheHype 🎙️

                              ⁠Legal Disclaimer for Machine Learning Made Simple⁠


                              16 min
                            • Machine Learning Made Simple Podcasts - Episode 17

                              Topic: Next-Level AI: Fine-Tuning with PEFT and LoRA for Industry Leaders

                              Summary:

                              • AI Agent Fine-Tuning: Strategies for fine-tuning AI agents to perform specific tasks akin to employees within an organization, enhancing role-specific functionalities.
                              • Collaborative AI Functionality: How multiple AI agents can work in tandem, simulating a cohesive workforce to streamline operations and increase efficiency.
                              • LoRA's Role in Efficiency: Exploring Low-Rank Adaptation (LoRA) as a method to facilitate fine-tuning with limited resources, such as 1 or 2 GPUs, by reducing the rank of training weights for more efficient model training.
                              • Productivity Enhancement: The impact of fine-tuned AI agents on workforce augmentation, demonstrating how AI can complement human personnel to boost productivity and drive innovation.

                                • Discover the innovative methods behind speeding up Large Language Model training, including the integration of PEFT and LoRA, and how they can be applied to create and manage efficient AI agents within the business environment.

                                If you enjoy this podcast, please:

                                • Subscribe to get notifications for new episodes.
                                • ⁠Follow me on LinkedIn⁠ for the latest in AI/ML papers and discussions.

                                  • About:

                                  Dive into the magic of machine learning with our podcast, where we unravel the mysteries in a language everyone can groove to! Ideal for the movers and shakers in the tech world – from top-tier execs shaping ML strategies to tech leads leading squads of MLEs. Whether you're an IT pro on the brink of a ML adventure or just someone itching to ride the ML wave, we've got your backstage pass to the world of ML hype! Tune in, turn up, and let's demystify machine learning together! 🚀✨ #MLGroove #DecodeTheHype 🎙️

                                  Legal Disclaimer for Machine Learning Made Simple


                                  19 min
                                • Machine Learning Made Simple Podcasts - Episode 16

                                  Topic: The Future of AI in Business: Understanding Transformer Architecture

                                  Summary:

                                  • Transformer Architecture: Unveiling the tech reshaping business.
                                  • Self-Attention Mechanisms: How they're the game-changer in AI.
                                  • LLM Fine-Tuning: Strategies for tailoring AI to your business needs.
                                  • Quantization: Optimizing AI models for efficiency and performance.

                                    • If you enjoy this podcast, please:

                                    • Subscribe to get notifications for new episodes.
                                    • Follow me on LinkedIn for the latest in AI/ML papers and discussions.


                                      • About:

                                      Dive into the magic of machine learning with our podcast, where we unravel the mysteries in a language everyone can groove to! Ideal for the movers and shakers in the tech world – from top-tier execs shaping ML strategies to tech leads leading squads of MLEs. Whether you're an IT pro on the brink of a ML adventure or just someone itching to ride the ML wave, we've got your backstage pass to the world of ML hype! Tune in, turn up, and let's demystify machine learning together! 🚀✨ #MLGroove #DecodeTheHype 🎙️

                                      Legal Disclaimer for Machine Learning Made Simple:

                                      Welcome to "Machine Learning Made Simple." The content provided through this Podcast is intended for general informational, educational, and entertainment purposes only. The views, opinions, and interpretations expressed by the host(s), guest(s), or contributors are their own and do not represent the views, policies, or positions of any company, organization, or institution with which they may be affiliated. Furthermore, despite referencing materials such as books, research papers, or other theoretical sources, all discussions are purely the personal opinions of the speakers and should be considered as such.

                                      Listeners should understand that while we strive to share insights, experiences, and perspectives that are informed and thought-provoking, these are not professional advisories. The content of the Podcast, therefore, should not be taken as legal, financial, technical, or professional advice. Any actions or decisions taken by listeners, inspired by the Podcast, are at their own discretion and risk. We encourage our audience to consult with qualified professionals before making any significant decisions based on the content of this Podcast.

                                      Machine Learning Made Simple and its contributors expressly disclaim any liability for the accuracy, timeliness, or completeness of the Podcast content. No warranty, expressed or implied, is provided in relation to the content, including, but not limited to, the application of the information discussed in real-life scenarios.

                                      By accessing, listening to, or participating in this Podcast, you agree to indemnify and hold harmless Machine Learning Made Simple, its producers, hosts, guests, and affiliates from any claims, damages, losses, liabilities, and expenses arising out of or in connection with the use of the information provided or opinions expressed in this Podcast.

                                      Please note that this disclaimer is subject to change without notice and should be reviewed periodically for any updates. Your continued engagement with the Podcast constitutes your agreement to be bound by the terms of this disclaimer.

                                      Thank you for listening to Machine Learning Made Simple. We hope our discussions offer valuable insights, though remember, applying any ideas or strategies discussed in the Podcast is done entirely at your own risk.

                                      28 min
                                    • Machine Learning Made Simple Podcasts - Episode 15

                                      Topic: The Layperson's Guide to AI's RNN and LSTM Technologies

                                      Summary:

                                      About: Dive into the magic of machine learning with our podcast, where we unravel the mysteries in a language everyone can groove to! Ideal for the movers and shakers in the tech world – from top-tier execs shaping ML strategies to tech leads leading squads of MLEs. Whether you're an IT pro on the brink of a ML adventure or just someone itching to ride the ML wave, we've got your backstage pass to the world of ML hype! Tune in, turn up, and let's demystify machine learning together! 🚀✨ #MLGroove #DecodeTheHype 🎙️

                                      Legal Disclaimer for Machine Learning Made Simple: Welcome to "Machine Learning Made Simple." The content provided through this Podcast is intended for general informational, educational, and entertainment purposes only. The views, opinions, and interpretations expressed by the host(s), guest(s), or contributors are their own and do not represent the views, policies, or positions of any company, organization, or institution with which they may be affiliated. Furthermore, despite referencing materials such as books, research papers, or other theoretical sources, all discussions are purely the personal opinions of the speakers and should be considered as such.

                                      Listeners should understand that while we strive to share insights, experiences, and perspectives that are informed and thought-provoking, these are not professional advisories. The content of the Podcast, therefore, should not be taken as legal, financial, technical, or professional advice. Any actions or decisions taken by listeners, inspired by the Podcast, are at their own discretion and risk. We encourage our audience to consult with qualified professionals before making any significant decisions based on the content of this Podcast.

                                      Machine Learning Made Simple and its contributors expressly disclaim any liability for the accuracy, timeliness, or completeness of the Podcast content. No warranty, expressed or implied, is provided in relation to the content, including, but not limited to, the application of the information discussed in real-life scenarios.

                                      By accessing, listening to, or participating in this Podcast, you agree to indemnify and hold harmless Machine Learning Made Simple, its producers, hosts, guests, and affiliates from any claims, damages, losses, liabilities, and expenses arising out of or in connection with the use of the information provided or opinions expressed in this Podcast.

                                      Please note that this disclaimer is subject to change without notice and should be reviewed periodically for any updates. Your continued engagement with the Podcast constitutes your agreement to be bound by the terms of this disclaimer.

                                      Thank you for listening to Machine Learning Made Simple. We hope our discussions offer valuable insights, though remember, applying any ideas or strategies discussed in the Podcast is done entirely at your own risk.

                                      24 min

                                    About Machine Learning Made Simple

                                    From the publisher's feed

                                    🎙️ Machine Learning Made Simple – The Podcast That Unpacks AI Like Never Before!