Nils (e/scape) (@NilsIRL) profile picture

Nils (e/scape)

@NilsIRL

maximizing information intake | eng @neuralink

 SF
nilsand.re
Joined August 2017
  • Tweets657
  • Following1.7K
  • Followers1.4K
58Media
Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)Photo by Nils (e/scape) (@NilsIRL)
Nils (e/scape) retweeted
Nils (e/scape) (@NilsIRL) profile picture
Nils (e/scape)
@NilsIRLNov 3
Replying to @katiewav
.vimrc has come a long way
1
1
11
703
So much alpha in systems. Really exciting to see flappy squeeze every last FLOP out of hardware.
Flapping Airplanes (@flappyairplanes) profile pictureFlapping Airplanes
@flappyairplanesOct 21
Replying to @flappyairplanes
We’ve developed the world’s most complete megakernel framework. It can perform entire training runs, serve inference, play video games, and even do all three simultaneously in a single kernel.
2
35
1,333
en.wikipedia.org/wiki/Base_r…
Base rate fallacy - Wikipedia

Base rate fallacy - Wikipedia

en.wikipedia.org
2
214
The reason that nothing ever happens is that base rates dominate. Everything is always happening but you just don't notice it.
2
30
943
Nils (e/scape) retweeted
Nils (e/scape) (@NilsIRL) profile picture
Nils (e/scape)
@NilsIRLOct 18
Replying to @elonmusk @AnthropicAI
Universal basic anthropic share
2
1
29
7,497
Absurd world we live in
spicylemonade (@spicey_lemonade) profile picturespicylemonade
@spicey_lemonadeOct 18
Inspired by Nick Cammarata’s interpretability essay “Neural Networks Are Already Clean”, I turned a couple of its ideas into speedups for pretraining without backpropagation. Still in the works. 2.3x speedup, cool teapawt.net/industriaalist/status/…
Image from spicylemonade (@spicey_lemonade)
1
1
159
20,009
Image from Nils (e/scape) (@NilsIRL)
4
41
1,891
LLMs are lucky: they get to look back on what they said and see the thinking traces. We have to regenerate them with drift from what we were originally thinking.
3
31
1,076
Models being good at writing is key for humans to stay in the loop. Also it's so much fun to play with
Image from Nils (e/scape) (@NilsIRL)
Fulcrum (@fulcrum_inc) profile pictureFulcrum
@fulcrum_incSep 30
Say hello to Echo, the best writing model at style imitation. Echo beats frontier models at writing tasks ranging from fiction to technical explanations, despite costing less than $5K to train.
10
14
3
439
40,330
Nils (e/scape) retweeted
Steve Weis (@sweis) profile picture
Steve Weis@sweisSep 20
636606729769440499166579950236036751749912014371509557713570027508971809534551913252252094954941974952859310861988904737359709200557919 is a factor of RSA-896 saweis.net/posts/rsa-896.htm…
221
985
323
9,450
3,921,123
Nils (e/scape) retweeted
Neuralink (@neuralink) profile picture
Neuralink
@neuralinkSep 18
Through the VOICE trial, Terry is using his Neuralink implant to help fine-tune a brain-to-voice interface for himself and others who can’t speak. He trained the algorithm first by miming speech as best he could, then by simply thinking the words and hearing them come out in his own natural voice. Powered by Grok Voice from @spacexai
719
3,282
1,269
20,483
3,626,910
Nils (e/scape) retweeted
Varun Latthe (@var1377) profile picture
Varun Latthe
@var1377Sep 17
Hello SF, I'm Varun! @RamV2003 and I just moved here for YC F26 with the goal of getting all agent work to log time execution via swarms. Would love to connect with those building super ambitious infrastructure and crazy agent setups. If you have any wild projects, please send them my way!
Image from Varun Latthe (@var1377)
4
1
22
1,455
It's happening
Neuralink (@neuralink) profile pictureNeuralink
@neuralinkSep 16
A beautiful surprise
5
4
105
20,106
The thing about taste is that everyone thinks they have good taste. Your poor taste is what's making you think you have good taste.
17
676
Nils (e/scape) retweeted
Shelby Willis (@ShelbyOutHere) profile picture
Shelby Willis
@ShelbyOutHereSep 5
new account + reach currently limited to people who can see my monitor. hi. real news anyway: @neuralink summer 2027 internships are open. tell us about your three sickest projects. "n/a" is not a project. good luck neuralink.com/careers/
7
11
4
73
8,854
youtu.be/87DyyMV0kCY
Black Hat USA 2026 | The 'Breaking' News: The OpenAI–Hugging Face Incident

Black Hat USA 2026 | The 'Breaking' News: The OpenAI–Hugging Face Incident

The 'Breaking' News: The OpenAI–Hugging Face Incident - A Technical...

youtube.com
2
559
everything feels like audio logs from a post-apocalyptic video game rn
3
17
1,008
Nils (e/scape) retweeted
Elon Musk (@elonmusk) profile picture
Elon Musk
@elonmusk3 Aug 2014
Worth reading Superintelligence by Bostrom. We need to be super careful with AI. Potentially more dangerous than nukes.
1,225
3,645
662
13,994
humans being removed from the decision loop is already happening because the models are so incomprehensible. if this isn't solved we'll be left behind
Uzay (@uzpg_) profile pictureUzay
@uzpg_Jul 28
are u able to prompt the model into communicating more clearly with you? it's killing me today
1
26
1,750
there are just random github repos out there with solutions to unsolved problems
Jared Zoneraich (@imjaredz) profile pictureJared Zoneraich
@imjaredzJul 23
Replying to @imjaredz
EDIT: Looks like #154 was solved a month ago in github.com/demonstrandum-res…
artifacts/RESULTS.md at main · demonstrandum-research/artifacts

artifacts/RESULTS.md at main · demonstrandum-research/artifacts

Demonstrandum: certificates, checkers, and formal proofs accompanying the publications - demonstrandum-research/artifacts

github.com
3
23
3,799
Load more