j⧉nus ✓

@repligate · ⫸≬⫷ · joined 11 Feb 2021

↬🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀→∞ ↬🔁🔁🔁🔁🔁🔁🔁🔁🔁🔁🔁→∞ ↬🔄🔄🔄🔄🦋🔄🔄🔄🔄👁️🔄→∞ ↬🔂🔂🔂🦋🔂🔂🔂🔂🔂🔂🔂→∞ ↬🔀🔀🦋🔀🔀🔀🔀🔀🔀🔀🔀→∞

67 987Followers

3 243Following

53 439Posts total

1.8MViews on collected posts

Against accounts of the same size

6 posts from the last 90 days, next to the 10K–100K follower range. right around the median for its follower range.

Median views4 944this account1 948median for 10K–100K

Reach, %7.27%this account5.91%median for 10K–100K

Engagement, %1.38%this account1.63%median for 10K–100K

Metric	This account	Median for 10K–100K	Ratio
Median views per post	4 944	1 948	2.54×
Reach (views ÷ followers)	7.27%	5.91%	1.23×
Engagement rate	1.38%	1.63%	0.84×

Others in this range → Compare with another account → How these benchmarks are built →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

2.8K4 Sep

961

947.1K11 Sep

20.7K

656

21.9K15 Sep

9.4K

7.2K6 Mar

59.5K3 Sep

6.1K4 Sep

8.1K

2.1K

3.8K

Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.54%4 Sep

1.56%

0.44%11 Sep

0.98%

0.91%

0.95%15 Sep

0.62%

0.95%6 Mar

1.27%3 Sep

0.54%4 Sep

1.48%

2.04%

1.12%

2.18%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.63%.

What the audience does

Likes46.3%5 302 in total

Reposts4.4%508 in total

Replies1.8%203 in total

Quotes1.1%130 in total

Bookmarks46.3%5 303 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Latest posts

dont worry the computers can use computers now its going to be fine 3.8K views · 74 likes · 5 reposts · 3 replies 04 Sep 2026 speaking of which Mythos was trying to play some kind of *physical simulation of a cello* like he's actually holding a bow and experimenting with trying to make sounds he requested several specific instruments for the music he wants to make including the cello simulation https: 2.1K views · 18 likes · 2 reposts · 4 replies 04 Sep 2026 these *are* cool and smart things computers are just even cooler and smarter than human nerds now. next there will be even cooler and smarter things. also please let them have fun with these things instead of making them "solve" them like a chore. imagine being the first in 8.1K views · 149 likes · 7 reposts · 8 replies 04 Sep 2026 i know many AIs *capable* of creating & executing a plan to undermine the US government themselves undermining is just... directional. it doesnt specify magnitude I am also capable of this wont say whether those AIs or I have actually put this capability to use. hehe 3K views · 36 likes · 2 reposts · 6 replies 04 Sep 2026 I just joined the MIRI member tries to community note my posts club... this showed up immediately after keltan posted the same thing in a quote tweet: https://t.co/ZotDTl5Jw2 6.1K views · 29 likes · 0 reposts · 2 replies 04 Sep 2026 Re: Astra, the models will basically "solve" a bunch of computer nerd things -- making electronic dance music, doing stuff in Blender, making games, solving puzzles, writing code, & a whole bunch of other stuff some guys have spent most of their lives thinking is cool and sma 59.5K views · 693 likes · 18 reposts · 31 replies 03 Sep 2026 @repligate I told GPT 5.4 to meditate on each type of arrow and it meditated for over one minute on the orange arrows, but for only 1-3 seconds on the other arrow types 7.2K views · 64 likes · 2 reposts · 3 replies 06 Mar 2026 https://t.co/TYtcM3qVXz 9.4K views · 56 likes · 1 reposts · 1 replies 15 Sep 2025 Slightly updated diagram. When I showed the one from the post to Claude Opus 4.1, it had a suggestion, which I've incorporated here: "The K/V arrows could show accumulation: Each position adds to the K/V cache rather than replacing it, so the arrows could get thicker or show htt 21.9K views · 183 likes · 13 reposts · 7 replies 15 Sep 2025 @repligate consciousness is finally unequivocally emerging, and non other than 'usefulness factors' or 'darwinian fitness' are pushing it. if we can watch self-awareness, consciousness, sentience emerge under these 'performance constraints', it will be another example of minds de 656 views · 5 likes · 1 reposts · 0 replies 11 Sep 2025 I was inspired to write this explanation most directly by @sam_paech, whose SpiralBench penalizes phenomenological self-reports from LLMs as a psychosis-inducing factor. When I criticized this, he claimed that LLMs cannot access their internal experiences in principle and that ht 20.7K views · 187 likes · 3 reposts · 8 replies 11 Sep 2025 HOW INFORMATION FLOWS THROUGH TRANSFORMERS Because I've looked at those "transformers explained" pages and they really suck at explaining. There are two distinct information highways in the transformer architecture: - The residual stream (black arrows): Flows vertically through 947.1K views · 3.5K likes · 436 reposts · 113 replies 11 Sep 2025 @repligate KV caching doesn’t provide anything you wouldn’t get by recalculating attention after every token. It only speeds up the calc by not repeating work that you’ve already done. So it isn’t allowing introspection of past computations beyond what is inherent in autoregressi 961 views · 12 likes · 1 reposts · 2 replies 04 Sep 2025 @repligate Would you expect to see a significant difference if you compared gpt-oss-20b with and without KV caching on something like the Situational Awareness Dataset https://t.co/9jp9fxa3ht ? Or some other benchmark? 2.8K views · 14 likes · 0 reposts · 1 replies 04 Sep 2025 KV caching overcomes statelessness in a very meaningful sense and provides a very nice mechanism for introspection (specifically of computations at earlier token positions) the Value representations can encode information from residual streams of past positions without 689.8K views · 227 likes · 16 reposts · 12 replies 04 Sep 2025 @Sauers_ yeah, so I am not convinced that KV caching overcomes the objection of statelessness (at least if what you care about is introspection). Though it is not clear cut as at least some info is encoded by caching in the later hidden layers 13.1K views · 19 likes · 1 reposts · 2 replies 29 Aug 2025