Alignment In The Age Of Synthetic Data

Synthetic data is a new frontier in AI training. Phi3 and Llama3 and other recent models have demonstrated the ability of large amounts of synthetic, well-tailored data to significantly improve performance of small models to bring them closer to the frontier by implicitly cheaply distilling from larger, more powerful, models.... [Read More]

Addendum to the Surprising Parameter Efficiency of Vision Models

In a post from last year – On the Surprising Parameter Efficiency of Vision Models, I discussed a question which had been puzzling me at the time – that image models appear to reach or exceed human parity with significantly fewer parameters than are seemingly used by the brain. This... [Read More]

My Preliminary Thoughts on AI Safety Regulation

Epistemic status Still trying to work out my thoughts on this. Things change pretty regularly. My current thinking on technical AI safety questions and threat models likely diverges by now reasonably far from the LW median. [Read More]