Tap a claim on the ring to put it at the centre.
← Training is no longer limited by data but by compute, because most of the data models learn from is now synthetic.
17 connected korrents · 16 moments from 10 Dec 2015 to 9 Oct 2026.
Everything filed under scaling laws
scaling laws
Everything filed under neural networks
neural networks
Everything filed under benchmarks
benchmarks
Everything filed under LLMs
LLMs
Everything filed under reinforcement learning
reinforcement learning 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: Training is no longer limited by data but by compute, because most of the data models learn from is now synthetic. Training is no longer limited by data but by compute, because most of the data models learn from is now synthetic. Last stated 7 months ago
23 Mar 2026
JH
Jensen Huang — holds since 2026-03-23 — tap for who they are
Same subject: AI will not run out of data, because there is already enough real-world data to build the simulators that generate more of it. — tap to centre the map on it AI will not run out of data, because there is already enough real-world data to build the simulators that generate more of it. Last stated a year ago
23 Jul 2025
DH
Demis Hassabis — holds since 2025-07-23 — tap for who they are
Same subject: When pre-training data is limited, it is better to train a smaller model for multiple epochs than to train a larger model on unique data once. — tap to centre the map on it When pre-training data is limited, it is better to train a smaller model for multiple epochs than to train a larger model on unique data once. Last stated 3 years ago
1 Dec 2023
SR
Sebastian Ruder — holds since 2023-12-01 — tap for who they are
Same subject: Data is becoming more valuable not only for training models but as a complement to cheaper intelligence. — tap to centre the map on it Data is becoming more valuable not only for training models but as a complement to cheaper intelligence. Last stated 2 weeks ago
9 Oct 2026
PC
Patrick Collison — holds since 2026-10-09 — tap for who they are
Same subject: The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out. — tap to centre the map on it The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out. Last stated a year ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
Same subject: AI's bottleneck was never its algorithms but how little data they were fed, which is why the field needed ImageNet rather than another model. — tap to centre the map on it AI's bottleneck was never its algorithms but how little data they were fed, which is why the field needed ImageNet rather than another model. Last stated 2 months ago
10 Aug 2026
FL
Fei-Fei Li — holds since 2026-08-10 — tap for who they are
Same subject: Available online text data is becoming a limiting factor for AI training, and repeating it yields diminishing returns. — tap to centre the map on it Available online text data is becoming a limiting factor for AI training, and repeating it yields diminishing returns. Last stated 3 years ago
19 Dec 2023
SR
Sebastian Ruder — holds since 2023-12-19 — tap for who they are
Same subject: Models are enormous only because the internet is bad: most of that capacity does memory work, and a better dataset would need far less of it. — tap to centre the map on it Models are enormous only because the internet is bad: most of that capacity does memory work, and a better dataset would need far less of it. Last stated a year ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
Same subject: Most future robotics training data will be robots' own autonomous attempts, the way most language-model data is now generated by the model itself. — tap to centre the map on it Most future robotics training data will be robots' own autonomous attempts, the way most language-model data is now generated by the model itself. Last stated 2 months ago
12 Aug 2026
CF
Chelsea Finn — holds since 2026-08-12 — tap for who they are
Same subject: A feedback loop that reinforces a behavior pulls it toward whatever improves the loop's own score. — tap to centre the map on it A feedback loop that reinforces a behavior pulls it toward whatever improves the loop's own score. Last stated 2 months ago
26 Aug 2026
HK
Henrik Karlsson — holds since 2026-08-26 — tap for who they are
Same subject: A physical AI company needs three AIs, not one — the agent, the simulator and the critic — turning deployment into a flywheel. — tap to centre the map on it A physical AI company needs three AIs, not one — the agent, the simulator and the critic — turning deployment into a flywheel. Last stated 3 months ago
3 Aug 2026
DD
Dmitri Dolgov — holds since 2026-08-03 — tap for who they are
Same subject: A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well. — tap to centre the map on it A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well. Last stated 11 months ago
17 Nov 2025
AK
Andrej Karpathy — holds since 2025-11-17 — tap for who they are
Same subject: A layer should learn a residual with reference to its own input rather than an unreferenced function, which is what makes great depth trainable. — tap to centre the map on it A layer should learn a residual with reference to its own input rather than an unreferenced function, which is what makes great depth trainable. Last stated 11 years ago
10 Dec 2015
JS
Jian Sun — holds since 2015-12-10 — tap for who they are
KH
Kaiming He — holds since 2015-12-10 — tap for who they are
Same subject: A small enough winning ticket learns faster than the network it was cut out of, and ends up more accurate than it. — tap to centre the map on it A small enough winning ticket learns faster than the network it was cut out of, and ends up more accurate than it. Last stated 9 years ago
9 Mar 2018
MC
Michael Carbin — holds since 2018-03-09 — tap for who they are
JF
Jonathan Frankle — holds since 2018-03-09 — tap for who they are
Same subject: A transduction model can rely entirely on self-attention for input and output representations without sequence-aligned RNNs or convolution. — tap to centre the map on it A transduction model can rely entirely on self-attention for input and output representations without sequence-aligned RNNs or convolution. Last stated 9 years ago
12 Jun 2017
IP
Illia Polosukhin — holds since 2017-06-12 — tap for who they are
JU
Jakob Uszkoreit — holds since 2017-06-12 — tap for who they are
NP
Niki Parmar — holds since 2017-06-12 — tap for who they are
AV
Ashish Vaswani — holds since 2017-06-12 — tap for who they are
NS
Noam Shazeer — holds since 2017-06-12 — tap for who they are
AG
Aidan N. Gomez — holds since 2017-06-12 — tap for who they are
+2
2 more
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 carmaker's claim to be the safest is mostly an artefact of comparing a new car against a fleet average twelve years old. — tap to centre the map on it A carmaker's claim to be the safest is mostly an artefact of comparing a new car against a fleet average twelve years old. Last stated 3 years ago
19 Dec 2023
PK
Philip Koopman — holds since 2023-12-19 — tap for who they are
Same subject: A company should define what great looks like for its own role rather than borrow another company's definition. — tap to centre the map on it A company should define what great looks like for its own role rather than borrow another company's definition. Last stated 2 months ago
9 Aug 2026
AW
Adam Ward — holds since 2026-08-09 — tap for who they are
At the centre
Training is no longer limited by data but by compute, because most of the data models learn from is now synthetic.
Holds Read this korrent →
AI will not run out of data, because there is already enough real-world data to build the simulators that generate more of it.
Holds
When pre-training data is limited, it is better to train a smaller model for multiple epochs than to train a larger model on unique data once.
Holds
Data is becoming more valuable not only for training models but as a complement to cheaper intelligence.
Holds
The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out.
Holds
AI's bottleneck was never its algorithms but how little data they were fed, which is why the field needed ImageNet rather than another model.
Holds
Available online text data is becoming a limiting factor for AI training, and repeating it yields diminishing returns.
Holds
Models are enormous only because the internet is bad: most of that capacity does memory work, and a better dataset would need far less of it.
Holds
Most future robotics training data will be robots' own autonomous attempts, the way most language-model data is now generated by the model itself.
Holds
A feedback loop that reinforces a behavior pulls it toward whatever improves the loop's own score.
Holds
A physical AI company needs three AIs, not one — the agent, the simulator and the critic — turning deployment into a flywheel.
Holds
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.
Holds
A layer should learn a residual with reference to its own input rather than an unreferenced function, which is what makes great depth trainable.
Holds
A small enough winning ticket learns faster than the network it was cut out of, and ends up more accurate than it.
Holds
A transduction model can rely entirely on self-attention for input and output representations without sequence-aligned RNNs or convolution.
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 carmaker's claim to be the safest is mostly an artefact of comparing a new car against a fleet average twelve years old.
Holds
A company should define what great looks like for its own role rather than borrow another company's definition.
Holds