Lossfunk Letters
Subscribe
Sign in
Home
Archive
About
Latest
Top
Discussions
First steps towards full stack functionalist theories of consciousness
Modeling attentional control and metacognitive evaluation, and taking phenomenology seriously, might point a way towards a full-stack theory of…
Nov 20
•
Sushrut Thorat
and
Paras Chopra
4
Subspace Crystallization in Grokking
Rank compression is a signature of Grokking and tracks the grokking onset.
Nov 5
•
Shlok Mehendale
5
1
October 2026
Steering AI Models’ Reasoning for Drug Repurposing
How sampling steering vectors from prior distributions stabilizes activations in reasoning models.
Oct 12
•
Lossfunk
and
Karush Suri
11
Your LLM won't believe you, but it WILL believe anyone "verified"
Models that shrug off a wrong user still flip for the same claim from a "verified source." We call this Authority Bias.
Oct 1
•
Abhinav Rajeev Kumar
11
2
1
September 2026
Traveling into the future
Reflections on 2 years of Lossfunk and the path ahead
Sep 7
•
Paras Chopra
34
12
3
August 2026
Diffusion models can see an illusion. Then they forget it before the picture comes out.
How denoising models quietly build human-like illusion representations deep inside the network, and throw the signal away before it ever reaches a…
Aug 3
•
Lossfunk
and
Gautam Ranka
3
2
June 2026
LLMs are blind. But somehow they can see.
How text-only models develop human-like maps of color, pitch, emotion, and taste, without ever experiencing any of them.
Jun 30
•
Lossfunk
8
1
1
The Metaprogramming Reflex: How the Best Coding Agents Survive Languages They've Never Seen
The agentic follow-up to EsoLang-Bench, now with tools, a workspace, and one strategy that separates the strong agents from the rest.
Jun 18
•
Aman Sharma
and
Lossfunk
4
2
May 2026
How to make progress on metaphysical puzzles in AI?
New paper from Lossfunk
May 21
•
Paras Chopra
5
3
1
Beyond Reward Design: Discovering RL Interfaces with LLMs
Jointly evolving RL observations and rewards with evolutionary LLM-guided search
May 11
•
Akshat Singh Jaswal
5
3
April 2026
Attributes of a great research question
Sharing our learnings from iterating on what good science is
Apr 29
•
Paras Chopra
25
2
10
Can AI models be conscious?
How can we tell?
Apr 21
•
Paras Chopra
12
1
3
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts