Tap a claim on the ring to put it at the centre.
← The special reasoning-trace delimiter tokens used by reasoning models…
17 connected korrents · 15 moments from 18 Mar 2024 to 21 Oct 2026.
Everything filed under LLMs
LLMs
Everything filed under Anthropic
Anthropic
Everything filed under open source
open source
Everything filed under OpenAI
OpenAI
Everything filed under AI and jobs
AI and jobs
Everything filed under robotics
robotics Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Read this korrent: The special reasoning-trace delimiter tokens used by reasoning models are cosmetic and not required for good reasoning performance. The special reasoning-trace delimiter tokens used by reasoning models are cosmetic and not required for good reasoning performance. Last stated 4 months ago
18 Jul 2026
SR
Sebastian Raschka — holds since 2026-07-18 — tap for who they are
Same subject: Tokens are not the true unit of AI inference, because better models yield more problem-solving per token. — tap to centre the map on it Tokens are not the true unit of AI inference, because better models yield more problem-solving per token. Last stated 2 months ago
15 Sept 2026
PG
Paul Graham — holds since 2026-09-15 — tap for who they are
Same subject: As labs sell finished white-collar work rather than reasoning traces, Chinese models will lose the ability to distil their way to parity. — tap to centre the map on it As labs sell finished white-collar work rather than reasoning traces, Chinese models will lose the ability to distil their way to parity. Last stated 8 months ago
13 Mar 2026
DP
Dylan Patel — holds since 2026-03-13 — tap for who they are
Same subject: Large language models do not reason formally: their performance collapses as a problem is made bigger, in the way a calculator's never does. — tap to centre the map on it Large language models do not reason formally: their performance collapses as a problem is made bigger, in the way a calculator's never does. Last stated 2 years ago
11 Oct 2024
GM
Gary Marcus — holds since 2024-10-11 — tap for who they are
Same subject: Reasoning problems, unlike general chat or writing tasks, can be automatically verified or labeled without human annotation. — tap to centre the map on it Reasoning problems, unlike general chat or writing tasks, can be automatically verified or labeled without human annotation. Last stated 2 years ago
27 Jan 2025
JA
Jay Alammar — holds since 2025-01-27 — tap for who they are
Same subject: Without good intuition for models, treat 100k tokens on smaller models and 200k on the frontier ones as the line past which the quality of your results is probably already degrading. — tap to centre the map on it Without good intuition for models, treat 100k tokens on smaller models and 200k on the frontier ones as the line past which the quality of your results is probably already degrading. Last stated 4 months ago
15 Jul 2026
DH
Dex Horthy — holds since 2026-07-15 — tap for who they are
Same subject: The aha moment for reasoning models will come from computer use and robotics rather than from scientific discovery, because those are the infinitely verifiable playgrounds. — tap to centre the map on it The aha moment for reasoning models will come from computer use and robotics rather than from scientific discovery, because those are the infinitely verifiable playgrounds. Last stated 2 years ago
3 Feb 2025
DP
Dylan Patel — holds since 2025-02-03 — tap for who they are
Same subject: Natural-language LLM mathematical reasoners are fundamentally limited by lacking machine-checkable semantics, which blocks autonomous verification needed for open-ended research. — tap to centre the map on it Natural-language LLM mathematical reasoners are fundamentally limited by lacking machine-checkable semantics, which blocks autonomous verification needed for open-ended research. Last stated 4 months ago
8 Jul 2026
TT
Terence Tao — holds since 2026-07-08 — tap for who they are
Same subject: Chain-of-thought explanations from language models can be biased because training does not explicitly reward faithful reasoning. — tap to centre the map on it Chain-of-thought explanations from language models can be biased because training does not explicitly reward faithful reasoning. Last stated 2 years ago
1 May 2025
LW
Lilian Weng — holds since 2025-05-01 — tap for who they are
Same subject: A PHP vulnerability found with GPT-5.6 Sol will cause havoc this week. — tap to centre the map on it A PHP vulnerability found with GPT-5.6 Sol will cause havoc this week. Last stated 2 weeks ago
21 Oct 2026
PS
Peter Steinberger — holds since 2026-10-21 — tap for who they are
Same subject: A startup should not begin life as a nonprofit and bolt a for-profit arm on later, whatever OpenAI's own history suggests. — tap to centre the map on it A startup should not begin life as a nonprofit and bolt a for-profit arm on later, whatever OpenAI's own history suggests. Last stated 3 years ago
18 Mar 2024
SA
Sam Altman — holds since 2024-03-18 — tap for who they are
Same subject: Advertising was a necessary phase for the internet but a momentary industry, and an AI people pay for is better because they know the answers are not influenced by advertisers. — tap to centre the map on it Advertising was a necessary phase for the internet but a momentary industry, and an AI people pay for is better because they know the answers are not influenced by advertisers. Last stated 3 years ago
18 Mar 2024
SA
Sam Altman — holds since 2024-03-18 — tap for who they are
Same subject: "I don't know" becomes a kind of superpower when you can work with Claude Code to figure things out. — tap to centre the map on it "I don't know" becomes a kind of superpower when you can work with Claude Code to figure things out. Last stated 8 months ago
19 Mar 2026
KP
Kris Puckett — holds since 2026-03-19 — tap for who they are
Same subject: A benchmark that ranks Claude Code last while it stays first in use is measuring the wrong thing, and has been for a year. — tap to centre the map on it A benchmark that ranks Claude Code last while it stays first in use is measuring the wrong thing, and has been for a year. Last stated 2 months ago
3 Sept 2026
DR
Dax Raad — holds since 2026-09-03 — tap for who they are
Same subject: A skill is a set of repeatable, specific instructions that augments what Claude Code can do. — tap to centre the map on it A skill is a set of repeatable, specific instructions that augments what Claude Code can do. Last stated 8 months ago
19 Mar 2026
KP
Kris Puckett — holds since 2026-03-19 — tap for who they are
Same subject: A business making hundreds of millions from an open source project owes it money or contribution, and can be refused its trademarks if it gives neither. — tap to centre the map on it A business making hundreds of millions from an open source project owes it money or contribution, and can be refused its trademarks if it gives neither. Last stated 2 years ago
26 Sept 2024
MM
Matt Mullenweg — holds since 2024-09-26 — tap for who they are
Same subject: A codec does not really exist until everyone can decode it. — tap to centre the map on it A codec does not really exist until everyone can decode it. Last stated 5 months ago
31 May 2026
JK
Jean-Baptiste Kempf — holds since 2026-05-31 — tap for who they are
Same subject: A company that will not accept a project’s terms is free to go and use a more permissive project instead. — tap to centre the map on it A company that will not accept a project’s terms is free to go and use a more permissive project instead. Last stated 2 years ago
26 Sept 2024
MM
Matt Mullenweg — holds since 2024-09-26 — tap for who they are
At the centre
The special reasoning-trace delimiter tokens used by reasoning models are cosmetic and not required for good reasoning performance.
Holds Read this korrent →
Tokens are not the true unit of AI inference, because better models yield more problem-solving per token.
Holds
As labs sell finished white-collar work rather than reasoning traces, Chinese models will lose the ability to distil their way to parity.
Holds
Large language models do not reason formally: their performance collapses as a problem is made bigger, in the way a calculator's never does.
Holds
Reasoning problems, unlike general chat or writing tasks, can be automatically verified or labeled without human annotation.
Holds
Without good intuition for models, treat 100k tokens on smaller models and 200k on the frontier ones as the line past which the quality of your results is probably already degrading.
Holds
The aha moment for reasoning models will come from computer use and robotics rather than from scientific discovery, because those are the infinitely verifiable playgrounds.
Holds
Natural-language LLM mathematical reasoners are fundamentally limited by lacking machine-checkable semantics, which blocks autonomous verification needed for open-ended research.
Holds
Chain-of-thought explanations from language models can be biased because training does not explicitly reward faithful reasoning.
Holds
A PHP vulnerability found with GPT-5.6 Sol will cause havoc this week.
Holds
A startup should not begin life as a nonprofit and bolt a for-profit arm on later, whatever OpenAI's own history suggests.
Holds
Advertising was a necessary phase for the internet but a momentary industry, and an AI people pay for is better because they know the answers are not influenced by advertisers.
Holds
"I don't know" becomes a kind of superpower when you can work with Claude Code to figure things out.
Holds
A benchmark that ranks Claude Code last while it stays first in use is measuring the wrong thing, and has been for a year.
Holds
A skill is a set of repeatable, specific instructions that augments what Claude Code can do.
Holds
A business making hundreds of millions from an open source project owes it money or contribution, and can be refused its trademarks if it gives neither.
Holds
A codec does not really exist until everyone can decode it.
Holds
A company that will not accept a project’s terms is free to go and use a more permissive project instead.
Holds