PhDing in AI safety, currently poking at computational sparsity in LLMs
Lucy Farnik
Karma: 84
I agree with a lot of this, we discuss the trickiness of measuring this properly in the paper (Appendix E.1) and I touched on it a bit in this post (the last bullet point in the last section). We did consider normalizing by the L2, ultimately we decided against that because the L2 indexes too heavily on the size of the majority of elements rather than indexing on the size of the largest elements, so it’s not really what we want. Fwiw I think normalizing by the L4 or the L_inf is more promising.
I agree it would be good for us to report more data on the pre-trained vs randomized thing specifically. I don’t really see that as a central claim of the paper so I didn’t prioritize putting a bunch of stuff for it in the appendices, but I might do a revision with more stats on that, and I really appreciate the suggestions.
[PAPER] Jacobian Sparse Autoencoders: Sparsify Computations, Not Just Activations
Great post! I’ve also experienced this build-up at many points in my life, this framework seems really useful.
One thing I’ve found particularly helpful for addressing this is freewriting. Just sit down every morning with your laptop and commit to writing 750 words. Don’t constrain what you write about, just let it be a stream of consciousness. You can write about your favorite type of pizza or whatever, but in practice you’ll probably get bored of shallow things like this pretty quickly and start diving into more meaningful things. You’ll then start noticing small annoyances that you weren’t consciously aware of before and you’ll gradually process them. And there will probably be a lot of them, so the cumulative effect of doing this consistently is really large.
Typical mind fallacy obviously applies here, freewriting has been life-changing for me but it might be less effective for you. But I’m guessing a large enough subset of LW readers is sufficiently similar to me that this is high-EV to try. Note that it can take a few days to start getting value out of this, I’d recommend committing to doing this everyday for 2 weeks. If after 2 weeks you don’t notice your emotional potential energy decreasing, it’s probably not a good fit for you.
Another thing that works somewhat well for me is just having some time without much external stimulation, e.g. going for regular walks without my phone and with earplugs. Again, it just lets your mind wander and eventually discover small things that have been bothering you and need to be processed.
Completely agree! As I said in the post
> There are tons of things I would’ve loved to do in this paper if there had been more engineering-hours.