korrents

← Data processing, filtering and quality, not architecture, is the…

On the map

7 connected korrents · 7 moments on record from 3 Feb 2025 to 19 Aug 2026.

Same subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subject Data processing, filteringand quality, notarchitecture, is the number… NL Nathan Lambert — holds since 2025-02-03 What sits in a model's context is far clearer to it than its training data, which is trillions of tokens stirred into a soup of parameters. What sits in a model'scontext is far clearer… JD Jeff Dean — holds since 2026-07-30 Large language models start in exactly the wrong place, because they try to get by without a goal and therefore without any sense of better or worse. Large language modelsstart in exactly the… RS Richard Sutton — holds since 2025-09-26 The language best suited to AI is simply the one the models have seen most of in training, which is why they do so well on JavaScript, TypeScript and Python. The language bestsuited to AI is simply… AH Anders Hejlsberg — holds since 2026-05-13 Language choice is a temporary artefact of training data; in the fullness of time every language will work equally well with AI. Language choice is atemporary artefact of… SY Steve Yegge — holds since 2026-03-11 Every machine-learning deployment that has paid off so far left a person making the decision, which is the only reason imperfect models were useful. Every machine-learningdeployment that has… CF Chelsea Finn — holds since 2026-08-12 Low-quality data makes a robot model worse unless you tell the model the data is low quality; labelled as such, the same data makes it better. Low-quality data makesa robot model worse… CF Chelsea Finn — holds since 2026-08-12 Editing your writing with a model pulls it towards one homogeneous style, far enough that you lose track of what your own voice was. Editing your writingwith a model pulls it… AO Addy Osmani — holds since 2026-08-19
same subject