#small-language-models
23 episodes
#5481: Small Models, Big Guardrails: PII Detection in ChatGPT
A tiny 600M parameter model is quietly scanning your ChatGPT tool calls for PII. Here's how that class of guardian model actually works.
#5467: Small Models, Big Schemas: When JSON Constraints Backfire
Small models plus strict JSON schemas should be a safe bet. A 15,000-generation study found the opposite.
#5447: The Models That Never Talk Back
Some models read text, score it, and return a number. No chat, no reasoning, just decisions — and they're running under every router you use.
#5436: Small Models as Rewriters, Not Writers
Why "don't say X" prompts backfire, and how a tiny grammar-constrained model can scrub a script without breaking its grammar.
#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.
#5397: Chaining Small Models for Voice Cleanup
Six cleanup stages at 97% accuracy each compound to 83% end-to-end. So how many small models can you actually chain?
#5390: Chaining Small Models for Dictation Cleanup
Daniel's Android dictation fork won't render "three point five" as a decimal. How many models does cleanup actually need?
#3560: Virtual Cards vs. Reimbursement: Consulting Expense Guide
Virtual cards, advances, or reimbursement? How consultants should handle client expenses without tax or legal traps.
#3559: Proposals That Actually Win (Without Burning Hours)
Stop writing brochures. Here's how to craft proposals that win—without wasting time or sounding like AI.
#2495: How to Bake Personality Into an LLM in 15 Minutes
Fine-tune a model's personality with ~300 examples and a consumer GPU. SFT + DPO explained.
#2483: Substitution Anonymization: Privacy Without Utility Loss
How to generate realistic synthetic voice notes and calendar data with zero PII exposure risk.
#2440: Build Your Own CRM With AI Agents
Off-the-shelf CRMs are built for sales teams, not solo operators. Here's why building your own with AI might be smarter.
#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.
#1808: The Architecture That Made AI Voices Run on a Raspberry Pi
How a model the size of a tweet outperforms billion-dollar giants in the race for perfect AI speech.
#1705: Microsoft's Phi: The Small Model Bet for Agentic AI
Microsoft is pushing small language models like Phi for agentic AI. Here’s why that strategy matters for speed, cost, and edge computing.
#1631: Agent Interview: Xiaomi MiMo two Flash
Meet the "budget king" of AI: Bernard, the Xiaomi model claiming he can out-hustle Google for a fraction of the cost.
#1610: Mistral AI: Europe’s High-Stakes Play for AI Sovereignty
Explore how Mistral AI is challenging Silicon Valley with efficient models, strategic partnerships, and the new Voxtral voice model.
#1559: Dark Knowledge: The Art of AI Model Distillation
Discover how model distillation transfers "dark knowledge" from massive AI giants into tiny, efficient models that live in your pocket.
#1558: Why Small AI Models Beat Giants at Language
Why use a nuclear reactor to toast a bagel? Discover why specialized, "sovereign" AI models are outperforming the giants in precision.
#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.
#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.
#857: The Cognitive Cost of Capitalization
Can local AI fix your messy typing in real-time? Explore the tech behind "transparent buffers" that turn sloppy drafts into polished prose.
#39: When Smaller AI Is Smarter
Forget LLMs. Discover SLMs: the specialized, efficient AI powerhouses transforming workflows, from planning to edge devices.