“You may need a bit of fine-tuning of your probes so that they work on each person's particular fine-tuned model”
EB
Eric Bigelow
"The Cognitive Revolution" ·
“Only 1.6% of Torquay 1930 its triples break this rule, and 0.5% of Lama's, against 9% and 13% for the two base models.”
LessWrong (Curated & Popular) · "Personas in a Pre-Internet Language Model" by David Africa
Play from 28:05
28:05
Audio streams from the publisher’s own feed. Verbatim does not host or copy it.
What was described
Fine-tuned a pre-1931 base model on period-matched supervised data (SFT) to improve preference coherence
Transitivity violations drop to 1.6% (vs 9% for base) and 0.5% for Lama
Said nearby on the same subject
“You need to go into settings in chat GPT and there's a little section called data controls. And then you need to turn off the beautifully named improve the models for everyone.”
DG
Daniel Gara
The Money Puzzle ·
“We asked inkling to fine tune itself using tinker. The model wrote its own fine tuning job, ran it and evaluated the result.”
♫
Said on Everyday AI Podcast - An AI and ChatGPT Podcast
“We have an algorithm that will determine the best model… for 99% of the tasks… we spend $15-$5 million in AI-related spend… frontier models are used only for supervision”
♫
Said on Prof G Markets
Moments like this, every Monday
The strongest new moments on your topics, in the speakers' own words. Free for three topics.
No card. No password. One-click unsubscribe.
Track a competitor?
Pro watches up to 25 names or phrases and emails you the morning after they are said.
See Pro, $29 a monthVerbatim quotes a short extract and links to the episode; the episode belongs to its publisher. Remove a show.