Intro topic: GrillsNews/Links:<ul><li>You can’t call yourself a senior until you’ve worked on a legacy project<ul><li><a href="https://www.infobip.com/developers/blog/seniors-working-on-a-legacy-project">https://www.infobip.com/developers/blog/seniors-working-on-a-legacy-project</a></li></ul></li><li>Recraft might be the most powerful AI image platform I’ve ever used — here’s why<ul><li><a href="https://www.tomsguide.com/ai/ai-image-video/recraft-might-be-the-most-powerful-ai-image-platform-ive-ever-used-heres-why">https://www.tomsguide.com/ai/ai-image-video/recraft-might-be-the-most-powerful-ai-image-platform-ive-ever-used-heres-why</a></li></ul></li><li>NASA has a list of 10 rules for software development<ul><li><a href="https://www.cs.otago.ac.nz/cosc345/resources/nasa-10-rules.htm">https://www.cs.otago.ac.nz/cosc345/resources/nasa-10-rules.htm</a></li></ul></li><li>AMD Radeon RX 9070 XT performance estimates leaked: 42% to 66% faster than Radeon RX 7900 GRE<ul><li><a href="https://www.tomshardware.com/tech-industry/amd-estimates-of-radeon-rx-9070-xt-performance-leaked-42-percent-66-percent-faster-than-radeon-rx-7900-gre">https://www.tomshardware.com/tech-industry/amd-estimates-of-radeon-rx-9070-xt-performance-leaked-42-percent-66-percent-faster-than-radeon-rx-7900-gre</a> </li></ul></li></ul>Book of the Show<ul><li>Patrick: <ul><li>The Player of Games (Ian M Banks)<ul><li><a href="https://a.co/d/1ZpUhGl">https://a.co/d/1ZpUhGl</a> (non-affiliate)</li></ul></li></ul></li><li>Jason: <ul><li>Basic Roleplaying Universal Game Engine<ul><li><a href="https://amzn.to/3ES4p5i">https://amzn.to/3ES4p5i</a></li></ul></li></ul></li></ul>Patreon Plug <a href="https://www.patreon.com/programmingthrowdown?ty=h">https://www.patreon.com/programmingthrowdown?ty=h</a>Tool of the Show<ul><li>Patrick: <ul><li>Pokemon Sword and Shield</li></ul></li><li>Jason: <ul><li>Features and Labels ( <a href="https://fal.ai">https://fal.ai</a> )</li></ul></li></ul>Topic: Reinforcement Learning<ul><li>Three types of AI<ul><li>Supervised Learning</li><li>Unsupervised Learning</li><li>Reinforcement Learning</li></ul></li><li>Online vs Offline RL</li><li>Optimization algorithms<ul><li>Value optimization<ul><li>SARSA</li><li>Q-Learning</li></ul></li><li>Policy optimization<ul><li>Policy Gradients</li><li>Actor-Critic</li><li>Proximal Policy Optimization</li></ul></li></ul></li><li>Value vs Policy Optimization<ul><li>Value optimization is more intuitive (Value loss)</li><li>Policy optimization is less intuitive at first (policy gradients)</li><li>Converting values to policies in deep learning is difficult</li></ul></li><li>Imitation Learning<ul><li>Supervised policy learning</li><li>Often used to bootstrap reinforcement learning</li></ul></li><li>Policy Evaluation<ul><li>Propensity scoring versus model-based</li></ul></li><li>Challenges to training RL model<ul><li>Two optimization loops<ul><li>Collecting feedback vs updating the model</li></ul></li><li>Difficult optimization target<ul><li>Policy evaluation</li></ul></li></ul></li><li>RLHF &  GRPO</li></ul>

 <a href="https://www.patreon.com/programmingthrowdown" rel="payment" title="★ Support this podcast on Patreon ★">★ Support this podcast on Patreon ★</a>

Podurama

Home

Trending

Personal

Your Podcasts

New Episodes

My Files

Favourite Episodes

Tags (Folders)

Recently Played

Bookmarks

Followed Playlists

180: Reinforcement Learning

Programming Throwdown

AI Chapters

AI Insights

Reviews