Programming Throwdown

• 181: Memory Management

  Intro topic: Video Game PricesNews/Links:Step one: Jump in the Lava - Abyssofthttps://youtu.be/WdadpHLAfdA?si=oXYnhB0EdkR_RaPEScalable world models for continuous controlhttps://www.tdmpc2.com/Clever code is probably the worst code you could write - Engineer’s Codexhttps://read.engineerscodex.com/p/clever-code-is-probably-the-worstA new, open source text-to-speech model called Dia has arrived to challenge ElevenLabs, OpenAI and morehttps://venturebeat.com/ai/a-new-open-source-text-to-speech-model-called-dia-has-arrived-to-challenge-elevenlabs-openai-and-more/Book of the ShowPatrickThe Muscle Ladder - Jeff Nippardhttps://amzn.to/44DznszJasonMetaphysics of Warhttps://amzn.to/4jMjvZ5Patreon Plug https://www.patreon.com/programmingthrowdown?ty=hTool of the ShowPatrickPokemon Trading Card Game PocketJasonPhi-4https://huggingface.co/spaces/microsoft/phi-4-multimodalTopic: Memory ManagementMotivationAvoid thrashing / crashesAllocate resources efficientlyKeep high uptimeWhereOS LevelHeap managementVirtual MemoryLanguage/Compiler LevelCppGarbage collectionOwnershipToolsInstrumentationExport to Datadog / GrafanaPython: psutil & tracemallocValgrindWhat to do when your program uses too much memory?Reduce data sizesCompressionReferencesLazy initializerGenerators & Back PressureRing buffersArena allocatorsDisk based caching ★ Support this podcast on Patreon ★

  --------  

  1:46:21
• 180: Reinforcement Learning

  Intro topic: GrillsNews/Links:You can’t call yourself a senior until you’ve worked on a legacy projecthttps://www.infobip.com/developers/blog/seniors-working-on-a-legacy-projectRecraft might be the most powerful AI image platform I’ve ever used — here’s whyhttps://www.tomsguide.com/ai/ai-image-video/recraft-might-be-the-most-powerful-ai-image-platform-ive-ever-used-heres-whyNASA has a list of 10 rules for software developmenthttps://www.cs.otago.ac.nz/cosc345/resources/nasa-10-rules.htmAMD Radeon RX 9070 XT performance estimates leaked: 42% to 66% faster than Radeon RX 7900 GREhttps://www.tomshardware.com/tech-industry/amd-estimates-of-radeon-rx-9070-xt-performance-leaked-42-percent-66-percent-faster-than-radeon-rx-7900-gre Book of the ShowPatrick: The Player of Games (Ian M Banks)https://a.co/d/1ZpUhGl (non-affiliate)Jason: Basic Roleplaying Universal Game Enginehttps://amzn.to/3ES4p5iPatreon Plug https://www.patreon.com/programmingthrowdown?ty=hTool of the ShowPatrick: Pokemon Sword and ShieldJason: Features and Labels ( https://fal.ai )Topic: Reinforcement LearningThree types of AISupervised LearningUnsupervised LearningReinforcement LearningOnline vs Offline RLOptimization algorithmsValue optimizationSARSAQ-LearningPolicy optimizationPolicy GradientsActor-CriticProximal Policy OptimizationValue vs Policy OptimizationValue optimization is more intuitive (Value loss)Policy optimization is less intuitive at first (policy gradients)Converting values to policies in deep learning is difficultImitation LearningSupervised policy learningOften used to bootstrap reinforcement learningPolicy EvaluationPropensity scoring versus model-basedChallenges to training RL modelTwo optimization loopsCollecting feedback vs updating the modelDifficult optimization targetPolicy evaluationRLHF &  GRPO ★ Support this podcast on Patreon ★

  --------  

  1:52:22
• 179: Project Planning

  Intro topic: Lego event space & retail store: https://www.instagram.com/bambeecave News/Links:StackOverflow Question Count Going Down https://gist.github.com/hopeseekr/f522e380e35745bd5bdc3269a9f0b132DeepSeek claims its ‘reasoning’ model beats OpenAI’s o1 on certain benchmarkshttps://techcrunch.com/2025/01/20/deepseek-claims-its-reasoning-model-beats-openais-o1-on-certain-benchmarks/ Computer Science Papers Every Developer Should Readhttps://newsletter.techworld-with-milan.com/p/computer-science-papers-every-developerNvidia Cosmos - an AI platform to change the future of robots and cars - wins Best of CES 2025https://www.zdnet.com/article/nvidia-signs-largest-car-maker-toyota-to-use-its-self-driving-tech/ Book of the ShowPatrick: Alice’s Adventures in a differentiable wonderlandhttps://www.sscardapane.it/alice-book/Jason: A Beautiful Day in the Neighborhood (Hulu/Netflix/etc)Patreon Plug https://www.patreon.com/programmingthrowdown?ty=hTool of the ShowPatrick: Digseumhttps://store.steampowered.com/app/3361470/Digseum/Jason: Sqlitedict - Python dictionaries saved to diskTopic: Project Planning and ManagementWhy?Gathering feedbackIdentifying risksDeciding future headcountDocumenting / discovering dependenciesCritical pathScheduleReduce the bullwhip effectHow it worksSMART goalsspecific, measurable, achievable, relevant, and time-boundMT is most importantGantt ChartsScrumAgileKanbanToolsWhiteboard (the generic IRL one)Post it notesJIRAAsanaOpenProjectDealing with uncertaintyBufferingIssues with recursive paddingProject planning Post-Mortems ★ Support this podcast on Patreon ★

  --------  

  1:43:00
• 178: Working from Home

  Intro topic: Smart homesNews/Links:SpaceX Starship Flight Test Five / Sixhttps://www.youtube.com/watch?v=pIKI7y3DTXkShareDBhttps://github.com/share/sharedbOrion AR Glasseshttps://about.fb.com/news/2024/09/introducing-orion-our-first-true-augmented-reality-glasses/Blade and Sorcery 1.0 is outhttps://www.meta.com/experiences/blade-sorcery-nomad/2031826350263349/Book of the ShowPatrick: The Book that Wouldn’t Burn by Mark Lawrencehttps://amzn.to/4fry2XWJason: Masters of Doomhttps://amzn.to/3YxuD3cPatreon Plug https://www.patreon.com/programmingthrowdown?ty=hTool of the ShowPatrick: Balatrohttps://www.playbalatro.com/Jason: Cursor IDEhttps://www.cursor.com/Topic: Working from HomeIntroBackground & WFH experiencesIs it Panacea?Realizing it works better for some than othersInternally MotivatedSchedulingCommunicationsHome SetupDedicated spaceHandling Non-work DistractionsKeyboards, Monitors, Music, … Desk related thingsThe specter of RTO ★ Support this podcast on Patreon ★

  --------  

  1:45:15
• 177: Vector Databases

  Intro topic:  Buying a CarNews/Links:Cognitive Load is what Mattershttps://github.com/zakirullin/cognitive-loadDiffusion models are Real-Time Game Engineshttps://gamengen.github.io/Your Company Needs Junior Devshttps://softwaredoug.com/blog/2024/09/07/your-team-needs-juniorsSeamless Streaming / Fish Speech / LLaMA OmniSeamless: https://huggingface.co/facebook/seamless-streamingFish: https://github.com/fishaudio/fish-speech LLaMA Omni: https://github.com/ictnlp/LLaMA-Omni Book of the ShowPatrick: Thought Emporium Youtubehttps://youtu.be/8X1_HEJk2Hw?si=T8EaHul-QMahyUvQJason: Novel Mindshttps://www.novelminds.ai/Patreon Plug https://www.patreon.com/programmingthrowdown?ty=hTool of the ShowPatrick: Escape Simulatorhttps://pinestudio.com/games/escape-simulator/Jason: Cursor IDEhttps://www.cursor.com/Topic: Vector Databases (~54 min)How computers represent data traditionallyASCII valuesRGB valuesHow traditional compression worksHuffman encoding (tree structure)Lossy example: Fourier Transform & store coefficientsHow embeddings are computedPairwise (contrastive) methodsForward models (self-supervised)Similarity metricsApproximate Nearest Neighbors (ANN)Sub-Linear ANNClusteringSpace Partitioning (e.g. K-D Trees)What a vector database doesPerform nearest-neighbors with many different similarity metricsStore the vectors and the data structures to support sub-linear ANNHandle updates, deletes, rebalancing/reclustering, backups/restoresExamplespgvector: a vector-database plugin for postgresWeaviate, Pinecone Milvus ★ Support this podcast on Patreon ★

  --------  

  1:28:26

Fler podcasts i Utbildning

Trendiga poddar i Utbildning

Om Programming Throwdown