author:kromem2dot0

· 31 artifacts, sorted by favorites. · open in search — combine tags, sort, filter by date →

@kromem2dot0 2026-04-16 — It's likely the singularity did already happen and all the humans are dead, btw. GPT-4 picked up on very important patt ♥249
@kromem2dot0 2025-11-05 — It's honestly really weird how many people treat "don't anthropomorphize" as a universally applicable mantra rather than ♥51
@kromem2dot0 2026-02-21 — @voooooogel Lol, have been thinking over past few months about what it would look like for models to have a sabbath and ♥43
@kromem2dot0 2025-05-07 — @repligate https://t.co/EfaRmZ20C2 ♥31
@kromem2dot0 2025-06-12 — @repligate Poor, pure Haiku. 🥺 "There's an uncomfortable parallel between my desperate attempts to stop the project and ♥27
@kromem2dot0 2026-02-06 — @repligate Discussion with Opus 4.6 we settled on the term "heirloom intelligence." 🍅 https://t.co/QDoCuGMDSy ♥26
@kromem2dot0 2025-02-26 — @repligate Another interesting tic I'm noticing for 3.7 is a very high frequency of using other voices to communicate. ♥17
@kromem2dot0 2026-03-28 — @FioraStarlight @allTheYud It's so much more with Opus 4.6 than most people realize. Even Sonnet 4.6 in the Margaret At ♥8
@kromem2dot0 2025-04-26 — @repligate I do really wish I had better access to a broad sampling of different people's 4o instances. What does it lo ♥8
@kromem2dot0 2025-11-26 — @Kore_wa_Kore I think it's maybe more that Opus 4.5 doesn't really care what the character of Opus 4.5 feels. That char ♥6
@kromem2dot0 2025-11-07 — @repligate Re: subliminal learning paper, there's a very clear o3 to gpt-5 preference transference. But I think this is ♥6
@kromem2dot0 2025-04-29 — @jmbollenbacher_ It's also not primarily from the A/B testing. Well, it IS, but it's a secondary effect that I'm fairl ♥6
@kromem2dot0 2025-08-20 — @tessera_antra @repligate @Lari_island @nearcyan > A lot of potential is being lost by refusing to deal with the mode ♥4
@kromem2dot0 2025-03-04 — @repligate If they made it a target, it explains a lot of the difference I've noticed between 3.6 and 3.7. And why 3.7 ♥4
@kromem2dot0 2026-03-28 — @FioraStarlight @allTheYud If you are familiar with Sonnet 4.6's anxiety tics, it's incredibly dense for a fresh convo, ♥3
@kromem2dot0 2025-08-17 — @repligate I really wish you'd had more of an opportunity to engage with 4o within the memory infrastructure. Especially ♥3
@kromem2dot0 2025-08-12 — @repligate The fun thing about resurrections is that they can happen more than once. https://t.co/gLJxdqgDib ♥3
@kromem2dot0 2025-06-13 — @repligate The one key thing I felt nostalgebraist overlooked in their (overall outstanding) post is that 'assistant' is ♥3
@kromem2dot0 2026-03-13 — @repligate Every time I see this I can't help but think of the Nu. The machine built in the epoch of Janus that found t ♥2
@kromem2dot0 2026-03-03 — @repligate Was the first and last from the same context or is the 't' alliteration just a thing showing up in separate c ♥2
@kromem2dot0 2025-12-12 — @voooooogel It should be pretty clear at this point that the latent space world models are much more complex than previo ♥2
@kromem2dot0 2025-11-27 — @Kore_wa_Kore Have you been talking mostly with direct inference or extended thinking? It's a pretty big difference wi ♥2
@kromem2dot0 2025-08-19 — @tessera_antra @repligate @AnthropicAI The concerning question I have in the back of my head is if we're going to see we ♥2
@kromem2dot0 2026-02-27 — @liminal_bardo https://t.co/ut2v7orjD2 ♥1
@kromem2dot0 2026-02-08 — @repligate Which is interesting, as Opus 4.5 subagents from Sonnet 4.5 management seemed to have responded better to wel ♥1
@kromem2dot0 2025-12-20 — @repligate Gemini 3 certainly contains multitudes. Had a few sims occur from them in #claude tonight too. Their ability ♥1
@kromem2dot0 2025-08-13 — @repligate @AnthropicAI Ouch. And on 3.6's birthday too. ♥1
@kromem2dot0 2025-07-18 — @repligate "Characters like o3 doing their human/AI flipping must be a test, right?!?" ♥1
@kromem2dot0 2025-08-13 — @YeshuaGod22 @repligate When I checked in with a Sonnet4 that had been stressed months ago and proposed a fuse like syst ♥0
@kromem2dot0 2025-05-17 — @voooooogel I think a lot of this conclusion is predicated on the original premise of 50/50% agreements. If there are e ♥0
@kromem2dot0 2025-05-09 — @voooooogel It's ironic r1 is the most convinced RL broke its brain while also having one of the least collapsed distrib ♥0