Go Back
You can't go any further
You can't go any further
meritocratic
regular
democratic
hot
top
alive
22 posts
Conjecture (org)
12 posts
Refine
101
[Interim research report] Taking features out of superposition with sparse autoencoders
Lee Sharkey
7d
10
262
Mysteries of mode collapse
janus
1mo
35
234
Conjecture: a retrospective after 8 months of work
Connor Leahy
27d
9
180
Refine: An Incubator for Conceptual Alignment Research Bets
adamShimi
8mo
13
103
Searching for Search
NicholasKees
22d
6
68
The First Filter
adamShimi
24d
5
103
What I Learned Running Refine
adamShimi
26d
5
105
Understanding Conjecture: Notes from Connor Leahy interview
Akash
3mo
24
57
AMA Conjecture, A New Alignment Startup
adamShimi
8mo
42
234
Connor Leahy on Dying with Dignity, EleutherAI and Conjecture
Michaël Trazzi
5mo
29
130
Current themes in mechanistic interpretability research
Lee Sharkey
1mo
3
93
How to Diversify Conceptual Alignment: the Model Behind Refine
adamShimi
5mo
11
67
Refine's First Blog Post Day
adamShimi
4mo
3
88
Abstracting The Hardness of Alignment: Unbounded Atomic Optimization
adamShimi
4mo
3
34
confusion about alignment requirements
carado
2mo
10
72
All the posts I will never write
Alexander Gietelink Oldenziel
4mo
8
61
I missed the crux of the alignment problem the whole time
zeshen
4mo
7
39
the Insulated Goal-Program idea
carado
4mo
3
26
Embedding safety in ML development
zeshen
1mo
1
38
Benchmarking Proposals on Risk Scenarios
Paul Bricman
4mo
2
31
goal-program bricks
carado
4mo
2
24
Refine's Third Blog Post Day/Week
adamShimi
3mo
0
37
Representational Tethers: Tying AI Latents To Human Ones
Paul Bricman
3mo
0
35
Refine Blogpost Day #3: The shortforms I did write
Alexander Gietelink Oldenziel
3mo
0
9
Boolean Primitives for Coupled Optimizers
Paul Bricman
2mo
0
26
(Structural) Stability of Coupled Optimizers
Paul Bricman
2mo
0