kegan 3 is less obvious when the social values are around free thinking and agency. mfw pretending to be kegan 4 is kegan 3 in sf
107438
Ulisse Mini✓
@MiniUlisse
unschooled autodidact interested in inner work / self development / understanding myself and others. technical staff @ https://t.co/1HqI3jcQqL
355 Following1.3K Followers4.8K Posts
- "It’s so hard to accept I’m not perfectly rational and capable of accruing infinite power"0012546
- harder and less sexy and far more important lol. extremes are nice bc they never happen I used to really relate to Grenade by Bruno Mars, listening to it now i see he has issues (that i also had/have) love a song -> grow -> see song as evidence of mental illness໊ @ilswyth“i would kill for you.” “i would die for you.” well i would heal for you. i’d unlearn my toxic patterns, face my wounds, and become a healthier person so my love doesn’t hurt you2025909
໊@ilswyth“i would kill for you.” “i would die for you.” well i would heal for you. i’d unlearn my toxic patterns, face my wounds, and become a healthier person so my love doesn’t hurt you21924.7K98.8K1.9M- weird thing here is that i expect nick would have predicted all these results & it’s mostly talking past each other, different meanings of “understand”, standards of evidence, etcUlisse Mini @MiniUlisse@nickcammarata's essay says InceptionV1 neuron 4e:55 fires on pointed things and not round ones. on the essay's own photos (top), that holds. on other ImageNet photos (bottom), all four combinations happen: round fluffy faces fire, pointy German shepherds stay silent 🧵105398
- Posts @nickcammarata@nickcammarata's essay says InceptionV1 neuron 4e:55 fires on pointed things and not round ones. on the essay's own photos (top), that holds. on other ImageNet photos (bottom), all four combinations happen: round fluffy faces fire, pointy German shepherds stay silent 🧵
Nick @nickcammaratainterpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works44565K - me/cfs is horrible! I’ve known adic a long time and it’s been awful to see her life get destroyed from this. i really hope we get large advances in biology soon!adic @adic_9late night phenomenology of a sensory crash (forced no stim rest) in me/cfs. body/brain shut down into a lower level of consciousness to recover from taxing experiences004253
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works8872.6K19.9K4.6M- Posts @MiniUlisseto conclude, if there's a crisp pointed vs round distinction here, I can't find it! I'd love to see if nick can predict this neuron better with his understanding now (as opposed to before) or figure out other good sharable "proofs/evidence of understanding"109194
- Posts @MiniUlisseexperimental details and more experiments: (got seriously nerdsniped) there's more i'm curious about, like ideas for leveraging LLMs to their full epistemic potential (infinite smart researchers, blinded however you want!) but that will have to wait. I've spent too long on this already 😅009202
- bullish on approaches like thisJake Ward @_jake_wardModels love reward hacking. Maybe we should just let them! In our new paper, we let models reward hack during RL but we dictate which parameters learn this behavior. When we disable these parameters for deployment, the model keeps its new capabilities but doesn't hack!007501
Jake Ward@_jake_wardModels love reward hacking. Maybe we should just let them! In our new paper, we let models reward hack during RL but we dictate which parameters learn this behavior. When we disable these parameters for deployment, the model keeps its new capabilities but doesn't hack!1052373.5K303K- it bothers me when people use unfalsifiable as a synonym for unscientific. the gap between ‘hard to get evidence either way’ and ‘impossible to get evidence’ is infinite oceans wide1015509
