Open Source
596 episodes · Page 3 of 25
#6213: Ten Weirdest AI Evals Ever Published
A pelican on a bicycle, a badger for Groundhog Day, and a benchmark where passing is the bad outcome — the strangest AI evals ever run.
#6212: Woolly Sedate Snail and Other Model Failures
Most of Hugging Face's 2.96M models sit unread. But the dust has a personality — snails, pirates, and a matchbox that resigns.
#6211: Attention Is Still Winning: The Hybrid Model Census
Mamba layers shipped — inside hybrids. A census of Qwen, Nemotron, Kimi, Granite, and why MiniMax went back to full attention.
#6207: Two WERs, One Model: Inside ASR Eval Harnesses
Same model, same audio, two different word error rates. The gap lives in the normaliser — and that's just layer two of three.
#6205: Where Should Your Notebook Actually Run?
Colab's free GPU comes with a 12-hour wall and hidden secrets in your saved outputs. Here's where notebooks should actually live.
#6201: Inside Cloudflare's Clef: A Language Model That Can't Speak
Cloudflare's Clef reads text, images, and video but outputs only probabilities — no words. Here's what's actually under the hood.
#6199: Running AI Experiments Before an Election
What MLflow, Braintrust and notebooks actually do when you have five days and one forecast to get right.
#6192: What Actually Happens When You Train a Model
Data in, model out — except the "data" is pairs, the split is the whole game, and one leaked speaker can fake your score.
#6190: When Your AI API Quietly Changes Under You
Same API name, different model behind it. How to tell when the thing you tested stops being the thing you're using.
#6187: Can Your Laptop Borrow Your Desktop's RAM?
Two gigabytes free on the laptop, sixty-four idle on the desktop next door. Can swap over the network actually help?
#6182: Running a Private APT Repo for Family Machines
Daniel wants his Debian packages on family machines without becoming their unpaid update service. Here's the modern way to do it.
#6181: Shipping Private Android Apps to Your Family
Daniel wants to ship private Android apps to non-technical relatives. F-Droid repos, Basic Auth, and the Google verification cloud on the horizon.
#6180: Mapping Code-Switching: 113 Pairs, No Community
Researchers count 113 code-switched language pairs — but a pair is just a label on a file, not a community of speakers.
#6172: Recipes vs. Regeneration: AI Photo Editing in 2026
Two ways AI edits photos now: models that write a readable recipe of slider values, and models that just repaint the whole frame.
#6168: Hebrish: Hebrew-Aware Dictation That Actually Works
Two new speech models fix English-Hebrew code-switching dictation — and the whole training run cost about fifty cents of GPU time.
#6165: Watching 20 Batch Jobs at Once
Batch APIs give you a discount, then hand you a babysitting job. Here's what actually watches a batch run — and what doesn't.
#6163: Forking a Keyboard for One Voice Feature
Daniel built a Hebrew-English dictation engine. Now he wants to fuse it into someone else's keyboard — without inheriting its entire future.
#6162: Why Dictated Words Rewrite Themselves
Hold-to-talk dictation shows dimmed text that silently changes when you release the key. Here's the architecture behind that rewrite.
#6161: Turning Words Into Sound: G2P, IPA, and the Hebrew Vowel Problem
Text-to-phonetics tools exist, but the neural ones guess wrong on exactly the words you care about — names, brands, and Hebrew.
#6148: The Tiny Hebrew Models Behind the Show's Voice
Dicta's 300M-parameter nikudizer, Phonikud's adaptors, and the ktiv male model hiding inside a vowel adder.
#6147: Hebrew AI: Ears, Mouth, and Brain
Open Hebrew speech recognition beats the paid services. Hebrew text-to-speech is a different story — and the language models are a third.
#6138: Hebrish on iPhone: MLX, Core ML, or Wait?
Hebrish spells makolet right on Android. Getting that Hebrew-English dictation onto an iPhone means picking between MLX, Core ML, and sherpa-onnx.
#6137: Two Ways a Chinese Open Model Gets Censored
Same weights, different landlord: how to tell a serving-time filter from censorship baked into the model itself.
#6136: How We Chose Our New Script Writer
Ten models audition to write the show. Seven fail the censorship gate — including the workhorse that wrote three episodes in four.