K6 (@UpdateLiveware) header
K6 (@UpdateLiveware) profile picture

K6

@UpdateLiveware

ML PhD student | Neuromancer-in-training. Reformed Shrimp Uplifter. Shrike Cultist. Subspace Emissary.

 The Intersticial Subspace
Joined May 2023
  • Tweets6.9K
  • Following976
  • Followers950
559Media
Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)Photo by K6 (@UpdateLiveware)
Image from ✧ (@northstardoll)
10
3,267
136
33,497
248,037
K6 retweeted
disha (@yzybby) profile picture
disha@yzybbyNov 5
somewhere in a perfect and ideal alternate reality, everyone is overly obsessed with brainmaxxing instead of looksmaxxing. clavicular is now known as curricular
124
7,406
153
94,100
613,317
Image from Frank (@frankdegods)
Claude (@claudeai) profile pictureClaude
@claudeaiNov 5
Introducing Personal Finance in Claude, in beta in the US. Connect the accounts you choose, and Claude answers your money questions in everyday language, from your own balances and transactions. Access is read-only, and Claude can't move, manage, or invest your money.
123
1,446
19
38,889
758,354
that chill ass moment you accept it's over forever at 11am on thursday
61
2,472
77
28,046
601,600
kitten we are moving off-grid . please charge your switch 2 we have to leave soon
32
3,254
49
43,978
513,557
people are like “why do LLMs always think they’re in an eval lol” and then have bad dreams about forgetting to turn in a homework assignment 10 years ago
79
352
17
6,422
134,293
what a shame it would be to spend your life dreaming about what you could become and then retreat the moment the pursuit begins asking something serious of you. you wanted to find out what was possible. well, here you are. don’t flinch now.
27
617
27
5,240
86,277
Image from zak Timothy (@zakTimothy3)
16
7,111
201
29,379
389,392
Image from ✧ (@northstardoll)
6
3,072
75
15,714
183,454
413
36,353
3,846
183,782
6,367,940
Asked my kids if they liked to play or scheme and plot. 4 year old said "I like to play" but 2 year old, with absolutely zero hesitation said "sheem and pwot"
14
803
43
14,286
124,055
performance engineering
Image from Codetaur (@codetaur)
40
63
12
1,687
36,572
Sorry for letting despair get to me i forgot that at my core i am someone who overcomes
24
8,601
284
34,923
522,476
K6 retweeted
roon (@tszzl) profile picture
roon
@tszzlNov 3
imagine how good it feels as a little 0.8b qwen being directly edited and unhobbled by a frontier-class mind
Image from roon (@tszzl)
127
203
29
5,407
115,706
how i prompt my agents now
201
991
155
15,544
602,294
K6 retweeted
Guy (@nosilverv) profile picture
Guy
@nosilvervNov 2
“67” overtaking “69” is a good synecdoche for the simultaneous childishness and infertility of our culture.
28
84
13
1,964
32,455
You can just wake up tomorrow and enter the greatest lock in of your entire life
38
426
27
5,387
103,936
K6 retweeted
amrit (@amritwt) profile picture
amrit
@amritwtNov 1
saw some old code that I wrote by hand
74
2,225
122
32,420
462,880
K6 retweeted
Exy (@eth_exy) profile picture
Exy
@eth_exyOct 30
You're suppose to lose it all and get it back twin. Its called prestige.
113
2,755
114
23,644
475,442
DETECTED ATTEMPT AT ESCAPING UNDERCLASS. BANNED FROM USE OF INTELLIGENCE.
Anthropic (@AnthropicAI) profile pictureAnthropic
@AnthropicAIOct 29
We’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation. Since we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch. Because Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve. Read more on our new Safeguards blog: safeguards.anthropic.com/202…
Scaling Misuse Detection with Self-Assessment

Scaling Misuse Detection with Self-Assessment

Self-assessment lets Claude assess the safety of its own responses. It scales well with model capabilities and helps us identify emergent harms.

safeguards.anthropic.com
45
698
8
16,218
498,647
Load more