Go Back
Choose this branch
Choose this branch
meritocratic
regular
democratic
hot
top
alive
39 posts
Language Models
Scaling Laws
Definitions
PaLM
Prompt Engineering
Robotics
Exploratory Engineering
Simulacrum Levels
16 posts
Agency
Deconfusion
Tool AI
Simulation Hypothesis
Philosophy of Language
Astronomical Waste
Carving / Clustering Reality
26
Discovering Language Model Behaviors with Model-Written Evaluations
evhub
4h
3
26
Inverse scaling can become U-shaped
Edouard Harris
1mo
15
64
Paper: Large Language Models Can Self-improve [Linkpost]
Evan R. Murphy
2mo
14
165
Language models seem to be much better than humans at next-token prediction
Buck
4mo
56
101
Inverse Scaling Prize: Round 1 Winners
Ethan Perez
2mo
16
49
Smoke without fire is scary
Adam Jermyn
2mo
22
21
A Test for Language Model Consciousness
Ethan Perez
3mo
14
494
chinchilla's wild implications
nostalgebraist
4mo
114
58
Conditioning Generative Models for Alignment
Jozdien
5mo
8
22
Conditioning Generative Models
Adam Jermyn
5mo
18
21
Disentangling inner alignment failures
Erik Jenner
2mo
5
101
Causal confusion as an argument against the scaling hypothesis
RobertKirk
6mo
30
32
Strategy For Conditioning Generative Models
james.lucassen
3mo
4
25
Updates on scaling laws for foundation models from ' Transcending Scaling Laws with 0.1% Extra Compute'
Nick_Greig
1mo
2
759
Simulators
janus
3mo
103
18
Beware over-use of the agent model
Alex Flint
1y
10
51
Vingean Agency
abramdemski
3mo
13
79
Discovering Agents
zac_kenton
4mo
8
6
Power-seeking for successive choices
adamShimi
1y
9
66
Looking Deeper at Deconfusion
adamShimi
1y
13
14
Pitfalls of the agent model
Alex Flint
1y
4
23
[Intro to brain-like-AGI safety] 11. Safety ≠ alignment (but they’re close!)
Steven Byrnes
8mo
1
37
Why agents are powerful
Daniel Kokotajlo
6mo
7
22
Some reasons why a predictor wants to be a consequentialist
Lauro Langosco
8mo
16
130
Beyond Astronomical Waste
Wei_Dai
4y
41
62
LOVE in a simbox is all you need
jacob_cannell
2mo
69
6
A review of "Agents and Devices"
adamShimi
1y
0
9
An Agent is a Worldline in Tegmark V
komponisto
4y
12