All of metawrong's Comments + Replies

Finetuned a SAE on deceptive/non deceptive reasoning traces from  Gemma 9b

If you generate synthetic deceptive trajectories, how can you be sure the SAE is going to generalise to 'real' deceptive trajectories? Also in those cases why do you need to use SAEs, can you use probes instead?

How does this explain the Decoy effect ^[1]?

1. ^{^}

   I am not sure how real and how well researched the 'decoy effect' is

So you would expect Claude Opus 3 to be harder to interpret than Claude Sonnet 3.5 ?

My intuition is that larger models of the same capability would exhibit less super-position and thus be easier to interpret?

> The news is good, and there are now seven shows in my tier 1

@Zvi Which are the other shows in your tier 1?

LASR (https://www.lasrlabs.org/) is giving a £11,000 stipend for a 13 week program, assuming 40h/week it works out to ~$27

It's difficult to produce a counterargument when the main argument is 'trust me bro'

The main argument seems to be (except for the all caps BUY STONKS NOW):

>It is just very obvious that the market is erring

This is not obvious to me

Thank you for doing this!

Few random questions, of course feel free to say as much or as little as you want:

Have you been personally to the Gaza strip (or the West Bank) ? If yes - what are your impressions? How do people live there? Is it common for regular (non-military) Jewish people to hang around those places? How common is for Palestinians to hang around outside of Gaza and the West Bank? How common is for Jewish and Palestinian people to mix in everyday life and interact?

I have just recently learned about the Gilat Shalit prisoner exchange (a single ... (read more)