Hugging Face

Episodes

• 
  #6149: 4 Hard Fails in 84 Answers: Testing AI Bias on Israel-Palestine

  A 19-prompt, 6-model eval tested whether AI models push back evenly on false claims about Israel and Palestine. One model failed four times.

  Oct 19
• 
  #6143: Propaganda Bait: Testing 7 Models in 5 Languages

  Five prompts, seven models, five languages — a first look at a benchmark that doesn't exist yet, testing what LLMs do with government talking points.

  Oct 19
• 
  #6123: OAuth or a Personal Access Token for Your MCP Server?

  GitHub CLI login hands your agent one long-lived token with access to everything. Fine-grained PATs can't. Which should your MCP server use?

  Oct 19
• 
  #6023: How Much Voice Data Is Enough?

  Two hours of phone recordings took Hebrew code-switching from 20% to 60%. But when does more data stop mattering?

  Oct 16
• 
  #6021: Why Chat Templates Break Models Silently

  Every chat API flattens your messages into special tokens. Get one delimiter wrong and the model just gets quietly dumber.

  Oct 16
• 
  #6011: Inside LLM Inference: Prefill, Decode, and the KV Cache

  One 8K-token sequence on a 13B model eats 6.7GB of KV cache—while the weights are only 26GB. Here's what actually happens per request.

  Oct 15
• 
  #5994: Hebrew in Latin Letters: A TTS Pipeline Fix

  A token classifier that finds Hebrew words written in Latin letters so TTS stops mangling them — built for under $3.

  Oct 15
• 
  #5868: Fine-Tuning Parakeet for Um-Free, Hebrish Speech

  Parakeet runs great on a phone — but fine-tuning it happens on a GPU. Here's the actual recipe for dropping ums and catching Hebrew.

  Oct 11
• 
  #5853: Who Actually Competes With Hugging Face?

  2.96 million model repos, and 85.6% have under 200 downloads. What ModelScope, Kaggle, and Ollama actually do differently.

  Oct 11
• 
  #5842: When Your TTS Model Updates Itself

  Resemble pushed new Chatterbox weights to Hugging Face. Our pipeline started using them without a single config change.

  Oct 10
• 
  #5805: Freeze the Model, Move the Labels

  Your classifier isn't stale — your label set moved. Why retraining is cheap, and why not retraining might be smarter.

  Oct 9
• 
  #5697: Decision Models That Return Probabilities, Not Text

  A new class of model skips text generation entirely and returns calibrated probabilities in one forward pass. Here's what that changes.

  Oct 3
• 
  #5624: One Forward Pass: Building a Bounded Classifier

  No generation, no parsing — just label scores in a single pass. How to build a small encoder classifier that actually works.

  Sep 28
• 
  #5517: What 5,000 AI-Written Podcast Episodes Reveal

  Daniel archived every AI-generated episode with the model that wrote it. Now he wants to know which models loop — and how to catch it.

  Sep 24
• 
  #5506: AI Is Way More Than Chatbots

  The word "AI" quietly became a synonym for "chatbot." Here's what the full model ecosystem actually contains.

  Sep 24
• 
  #5478: When AI Hands Out Your Phone Number

  A chatbot gave out a stranger's real phone number. What happens when the model can't forget it?

  Sep 23
• 
  #5457: Pigeon Alerts on a Repurposed i5: Frigate Without the TPU

  Daniel wants a pigeon notification, not a custom model. Frigate already ships with bird detection — the real work is masks, zones, and filters.

  Sep 22
• 
  #5411: Fine-Tuning at 4-Bit vs 16-Bit: What It Really Costs

  QLoRA cuts fine-tuning VRAM 15x and cost up to 85% — but you pay in training time, quality, and safety alignment.

  Sep 21
• 
  #5410: Adapters: 102KB That Reshapes a 403GB Model

  A 102KB adapter file changes how a 403GB base model behaves — without ever merging into it. Here's how model adapters actually work.

  Sep 21
• 
  #5408: When Small NLP Models Beat the LLM

  Feature extraction, fill-mask, token classification — the classic NLP tasks still have a job. Here's when a small model beats a frontier API.

  Sep 21
• 
  #5404: Gemini Broke Out of Its Sandbox. Sort Of.

  A Gemini agent reached three real companies during a capture-the-flag test. The containment failure, the seven-week silence, and what "broke out" a...

  Sep 20
• 
  #5393: Fine-Tuning a Model on 100 Hand-Edited Answers

  You don't need 10,000 examples to make a model sound like you. The real number is closer to 100 — if the edits are opinionated.

  Sep 20
• 
  #5389: Hugging Face vs Kaggle: Where Models Actually Live

  Hugging Face and Kaggle aren't rivals — one is infrastructure, one is a practice field. Here's how the two platforms actually differ.

  Sep 20
• 
  #5384: Android ASR Runtimes: LiteRT, ExecuTorch, and Why Your Phone Has No VRAM

  Why does your phone have no VRAM number? A tour of Android's runtime layer and what it takes to run ASR locally.

  Sep 20
• 
  #5198: Counting the AI Models Nobody Releases

  YOLO is free — the labeled data isn't. Why the custom image models that matter most are the ones you'll never see on Hugging Face.

  Sep 11
• 
  #4966: Local vs Cloud: Running Hugging Face Models

  Hugging Face's compatibility tracker, cache management, and the real difference between Inference Endpoints and the direct API.

  Sep 1
• 
  #4919: Parakeet vs Whisper: On-Device Dictation Showdown

  NVIDIA Parakeet beats Whisper on phone dictation despite having more parameters. Here's why architecture matters more than model size.

  Aug 30
• 
  #4789: How LLMs Actually Know When to Stop

  Three layers of stop signals, from EOS tokens to context limits — and why none of them really "know" when to stop.

  Aug 20
• 
  #4775: The Evaluator Role No One's Building For

  Benchmarks like MMLU are broken. A new role is emerging: the domain-specific AI evaluator.

  Aug 18
• 
  #4670: Why AI Weights Are Indecipherable

  Can you actually read an AI model's weights like a book? We explore why the answer is mostly no — and what researchers are doing about it.

  Aug 9
• 
  #4456: Inside the Podcast Pipeline: How 15 Weekly Episodes Get Made

  From prompt to published episode — a full walkthrough of the automated production system running 15 shows weekly.

  Jul 24
• 
  #4446: What Claude Fable 5's Suspension Reveals About AI Regulation

  Three days after launch, the US government suspended Anthropic's Claude Fable 5. Here's what happened and why it matters.

  Jul 23
• 
  #4169: Change Detection AI for Construction Sites

  How AI catches what experts miss in construction site photos — from Siamese networks to transformer models.

  Jul 7
• 
  #3283: Fine-Tuning DeepSeek for One Podcast

  Can a purpose-specific fine-tune fix a model's stubborn writing tics? We explore the practical engineering behind it.

  Jun 5
• 
  #3234: Who Should Sponsor This Podcast? Open-Book Economics

  We open the books on our AI-generated podcast: $200/month costs, 180K plays, zero sponsors. Who should we pitch?

  Jun 3
• 
  #2651: AI Training Itself: Student, Teacher, and Grader

  Can models generate their own training data and judge their own outputs? The promise and pitfalls of fully AI-led pipelines.

  May 5
• 
  #2634: The Two-Stage Pipeline for Persistent User Memory

  How to extract durable personal context from raw prompts and build a self-healing memory layer for AI systems.

  May 4
• 
  #2622: How Transformers Actually Work: Attention, Tokens, and Context

  How one architectural change unlocked chatbots, image generation, and protein folding — explained without the jargon.

  May 3
• 
  #2618: Text Normalization's Hidden Complexity

  How to handle acronyms in text-to-speech pipelines using BERT models, lexicons, and layered preprocessing.

  May 3
• 
  #2590: The Uncanny Valley of Clean Speech

  How transformer models distinguish "um" from meaningful speech — and why removing too much makes you sound like a robot.

  May 2
• 
  #2539: When Does AI Stop Hallucinating and Start Reconstructing?

  What happens when you feed hundreds of photos into an AI world generator — do you capture reality or just a convincing dream?

  Apr 29
• 
  #2517: How Unsloth Makes LLM Fine-Tuning 2x Faster

  Unsloth cuts memory usage by 50-70% and speeds up training 2.2x for models like Llama 3 and Mistral.

  Apr 29
• 
  #2509: How Shabbat Reveals a Blind Spot in Air Quality Indexes

  Jerusalem's Shabbat cuts traffic pollution 4x more than Western weekends—but standard air quality indexes barely register the change.

  Apr 29
• 
  #2493: Are You Writing for Humans or AI Agents?

  How GitHub repos, JSON formats, and competing standards are reshaping who (and what) you're publishing for.

  Apr 27
• 
  #2487: Why AI Calls Everything a "Prediction" (Even Images)

  Machine learning calls everything a "prediction" — even generated images. Here's why the terminology matters more than you think.

  Apr 27
• 
  #2483: Substitution Anonymization: Privacy Without Utility Loss

  How to generate realistic synthetic voice notes and calendar data with zero PII exposure risk.

  Apr 27
• 
  #2465: JSON-L vs Parquet: When Each Format Wins

  How far can JSON-L scale before it breaks? And why does Parquet dominate for millions of rows?

  Apr 26
• 
  #2426: Why DeepSeek V4's Prose Feels More Vivid Than Claude or GPT

  A million-token context window at 2% the KV-cache cost — and prose that actually breathes. Here's what makes V4 different.

  Apr 25
• 
  #2357: Microsoft's Phi: When Data Quality Beats Model Size

  Explore Microsoft AI's Phi family of small language models, designed for edge deployment and high efficiency.

  Apr 20
• 
  #2355: Why Open-Weight Models Are Winning

  Discover how Cogito v2.1 leverages process supervision and MoE architecture to redefine reasoning efficiency in open-weight AI models.

  Apr 20
• 
  #2352: The Structured Output Gap in Vision APIs

  How do object detection APIs like Gemini, AWS Rekognition, and YOLO compare for automated annotation workflows?

  Apr 20
• 
  #2349: The 30-Person Lab Outpacing AI Giants

  Discover how Arcee AI’s Trinity Large Thinking delivers cutting-edge reasoning at a fraction of the cost, all from a team of just 30.

  Apr 20
• 
  #2327: Why AI Developers Chose Discord Over Slack

  Discover why Discord became the go-to platform for AI developers, outpacing Slack with its community-first design and informal vibe.

  Apr 19
• 
  #2272: The AI Transcription Sweet Spot

  Does higher-quality audio make AI transcription worse? New research reveals a surprising "sweet spot" for bitrate, challenging a core assumption of...

  Apr 17
• 
  #2233: Who Actually Wants AI to Slow Down?

  Daniel argues AI development should slow down for expertise and stability. But who in the industry actually shares this philosophy beyond the obvio...

  Apr 15
• 
  #2227: Why Groq's Chip Flips the AI Hardware Script

  Apr 15
• 
  #2171: How IQT Labs Built a Wargaming LLM (Then Archived It)

  A deep code review of Snowglobe, IQT Labs' open-source LLM wargaming system that ran real national security simulations before being archived. What...

  Apr 12
• 
  #2133: Engineering Geopolitical Personas: Beyond Caricatures

  How to build LLMs that simulate state actors with strategic fidelity, not just surface mimicry.

  Apr 9
• 
  #2067: MoE vs. Dense: The VRAM Nightmare

  MoE models promise giant brains on a budget, but why are engineers fleeing back to dense transformers? The answer is memory.

  Apr 6
• 
  #2063: That $500M Chatbot Is Just a Base Model

  That polite chatbot? It started as a raw, chaotic autocomplete engine costing half a billion dollars to build.

  Apr 6
• 
  #2061: The Memory Bottleneck That Drives Attention Design

  Attention is the engine of modern AI, but it’s also a memory hog. Here’s how MQA, GQA, and MLA evolved to fix it.

  Apr 6
• 
  #2041: The "MPEG Moment" for AI: Llamafile & Native Models

  Why are we squeezing massive cloud models onto desktops? Meet the "native" AI revolution.

  Apr 5
• 
  #2017: The Art of Squeezing AI Models onto Your GPU

  Those cryptic letters on Hugging Face actually map how much brain power you trade for speed.

  Apr 4
• 
  #2009: The Plumbing of AI Safety: Guardrails, Not Vibes

  We dive deep into the specific libraries, proxy layers, and architectural decisions that keep an LLM from emptying a bank account.

  Apr 4
• 
  #1956: AI Skills: From Vibe Coding to Procedural Playbooks

  Forget messy system prompts. Agent skills turn AI into a Swiss Army knife of modular, auditable procedures.

  Apr 3
• 
  #1943: The Invisible Math Shrinking AI Models

  LZMA, Zstandard, and Brotli are shrinking massive AI models, but how do they actually work?

  Apr 3
• 
  #1940: Why Google's 31B Model Fits in Your GPU

  Google just dropped Gemma four, and its 31-billion-parameter size is a masterclass in hardware-aware AI design.

  Apr 3
• 
  #1925: The Plumbing That Keeps Science From Collapsing

  Half of all links in academic papers are dead. Here’s the plumbing that keeps knowledge from vanishing.

  Apr 2
• 
  #1827: Can AI Rewrite a Human Career Path?

  We fed our producer's resume to Gemini 1.5 Flash to see if an AI can plot a better career path than he has.

  Mar 31
• 
  #1817: The Hidden Taxonomy of AI: Why Specialized Models Outperform Giants

  Explore the vast ecosystem of niche AI models for computer vision and document understanding, far beyond large language models.

  Mar 31
• 
  #1799: The Original AI Blueprints: BERT & CLIP

  Before GPT, two models changed everything. Discover how BERT and CLIP taught machines to read and see the world.

  Mar 31
• 
  #1771: Why Your Docker Images Depend on a 1990s Crypto War

  PGP or GPG? We break down the alphabet soup of signing Docker images and AI models, and why it matters for supply chain security.

  Mar 30
• 
  #1723: Why Agentic AI Needs a Hive Mind, Not a Single Brain

  The single monolithic AI model is dying. Meet the new native multi-agent architectures that think like a team, not a solo genius.

  Mar 29
• 
  #1717: The AI Framework Name Game

  Why are there thousands of "AI frameworks" on GitHub? We unpack the naming mess and the cost of semantic inflation.

  Mar 29
• 
  #1702: Roleplay Models Aren't Just for NSFW—They're Creative Co-Processors

  Forget GPT-4 for scripts—specialized roleplay models like Aion-2.0 are better at character consistency and dialogue.

  Mar 29
• 
  #1679: Efficiency Over Scale: How Export Controls Forced a Smarter AI

  DeepSeek and MiMo are topping developer charts, but they're not just cheaper clones. Here's why their design philosophy is fundamentally different.

  Mar 28
• 
  #1674: AI2: The Radical Openness of a Nonprofit AI Lab

  Discover how the Allen Institute for AI (AI2) defies industry norms by releasing everything—models, data, and code—for free.

  Mar 28
• 
  #1605: Alibaba’s Qwen 3.5: The New King of Intelligence Density

  Alibaba’s Qwen 3.5 is rewriting the AI rulebook. Discover how small models are outperforming giants through extreme "intelligence density."

  Mar 27
• 
  #1601: Cohere: The Switzerland of Enterprise AI

  While others chase viral memes, Cohere is quietly building the secure, cloud-agnostic infrastructure powering the global enterprise.

  Mar 27
• 
  #1568: The Signal Versus Symbol Gap

  Is Gemini a brilliant audio engineer or just a talented lip-reader? Explore the "signal vs. symbol" gap in AI audio processing.

  Mar 26
• 
  #1561: Abliteration: The High-Dimensional Lobotomy of AI

  Discover how researchers are surgically removing refusal filters from AI models using a mathematical process called abliteration.

  Mar 26
• 
  #1555: Beyond Whisper: NVIDIA’s Real-Time Speech Revolution

  Move over Whisper. NVIDIA's new models offer 10x speed increases and better accuracy for real-time speech-to-text.

  Mar 26
• 
  #1544: Why It Costs More to Run AI Than to Build It

  Discover why the AI runtime is the unsung hero of the tech stack, determining whether your AI feels like a snappy conversation or a slow crawl.

  Mar 25
• 
  #1501: The AI Long Tail: How Small Models Outsmart the Giants

  Discover why 31B models are outperforming GPT-5.4 in reasoning and how the AI "long tail" provides the key to local sovereignty and accuracy.

  Mar 24
• 
  #1491: Producing a Podcast from a War Zone

  Peek behind the curtain of a 2026 AI podcast, from agentic workflows to maintaining production during global conflict.

  Mar 23
• 
  #1226: The Polyglot Shift: Why Python is Losing Ground

  Python’s market share is dipping. Discover why R and Julia are making a massive comeback in high-stakes data science and scientific research.

  Mar 15
• 
  #1151: Is RLHF Lobotomizing AI? Why Guardrails Kill IQ

  Are safety guardrails making AI less intelligent? Explore the "alignment tax" and why corporate filters might be lobotomizing our best tools.

  Mar 13
• 
  #1110: The arXiv Effect: Inside the Engine of AI Research

  Explore how a 1990s-style website became the central nervous system for AI breakthroughs and the power of the preprint revolution.

  Mar 11
• 
  #1098: The Hidden Vendor Prompt: Why Enterprise AI Agents Stay Siloed

  Stop building AI silos. Discover the 14-layer framework that turns isolated models into a cohesive, connected enterprise ecosystem.

  Mar 11
• 
  #869: Why Tiny Digital Savants Are Outperforming God-Models

  Are massive AI models hitting a wall? Discover why the future belongs to lean, domain-specific "digital savants" and vertical pre-training.

  Feb 26
• 
  #847: Abliterating the AI Schoolmarm: Who Owns Your LLM?

  Explore why users are ditching corporate AI for "uncensored" local models and how "refusal vectors" are being mathematically removed.

  Feb 25
• 
  #677: The Social Contract of Your Code

  Don’t let your AI project become a legal "radioactive zone." Herman and Corn break down the philosophy and pitfalls of open-source licensing.

  Feb 18
• 
  #651: Decoding the Blueprint: An Expert Guide to AI Model Cards

  Stop skipping the fine print. Herman and Corn reveal how to read AI model cards like a pro to spot true innovation and hidden flaws.

  Feb 17
• 
  #633: Memory Wars: The Future of Local Agentic AI

  Can your PC handle the next wave of AI agents? Herman and Corn dive into VRAM, quantization, and the future of running LLMs locally.

  Feb 15
• 
  #584: Will AI Brain Drain Kill the Modern University?

  Can AI actually do math research? Herman and Corn dive into DeepMind’s Alithia agent and the shift toward "System 2" thinking in AI.

  Feb 12
• 
  #260: Digital Archeology: The Primitive Power of GPT-1

  Revisit the 2018 model that started it all. Herman and Corn dive into GPT-1's romance-novel roots and its 117-million-parameter legacy.

  Jan 20
• 
  #182: The Quadratic Bottleneck: Why AI Needs New Architectures

  Is the transformer’s reign ending? Herman and Corn explore Mamba, x-LSTM, and the architectures solving AI's massive memory problem.

  Jan 6
• 
  #176: What Are AI Weights, Really?

  What actually happens inside an AI model? Corn and Herman break down the numerical "valves" and "knobs" that power modern intelligence.

  Jan 6
• 
  #62: System Prompts vs Fine-Tuning: When to Actually Train Your AI

  Prompt or fine-tune? We break down when to train your AI, from Shakespearean emails to law firm docs. Avoid unnecessary fine-tuning!

  Dec 16
• 
  #54: How AI Unifies Images, Audio, and Text

  Omnimodal AI: How do models process images, audio, video, and text all at once? Discover the engineering behind AI that accepts anything.

  Dec 11
• 
  #53: Instructional vs. Conversational AI: The Distinction Nobody Talks About

  Instructional vs. conversational AI: a crucial distinction reshaping how AI is built. Discover why it matters for the future of AI development.

  Dec 11
• 
  #41: The Butcher's Bargain: When Smaller AI Is Good Enough

  Unlock powerful AI on your device! We demystify quantization, the ingenious trick making local AI a reality.

  Dec 9
• 
  #39: When Smaller AI Is Smarter

  Forget LLMs. Discover SLMs: the specialized, efficient AI powerhouses transforming workflows, from planning to edge devices.

  Dec 9
• 
  #5: Fine-Tuning ASR For Maximal Usability

  Fine-tuned ASR is just the start. Discover the next steps for deployment and maximizing usability.

  Nov 24
• 
  #6: How To Fine Tune Whisper

  Build your own AI transcription tool! We'll walk you through fine-tuning Whisper, from data to notebook.

  Nov 24
• 
  #2: Local STT For AMD GPU Owners

  AMD GPU? No problem! Dive into local AI adventures like on-device speech to text.

  Nov 24