Home
About
CV
Mohammed Abu Baker (Shahoyi)
Doing research in AI Safety
Categories
All
(3)
AI Safety
(3)
RL
(1)
Sleeper Agents
(1)
Technical
(3)
ARENA 6.0 Capstone: Model Organism of Encoded Reasoning
AI Safety
Technical
RL
TL;DR For our capstone project in ARENA 6.0 (Sep 2025), we tried to create a model organism of encoded reasoning from Qwen3-4B using RL. Specifically, we tried to make the…
Nov 5, 2025
Luca Baroni, Mo Baker
AI Safety Mindmap
AI Safety
Technical
AI safety is a broad with many suibfields each prioritising and focusing on different asepcts of making AI safe. Therefore, I decided to crate a mindmap to hopefully capture…
May 20, 2025
Mo Baker
Changes in Attention Patterns of Data Poisoned Sleeper Agent LLMs
Sleeper Agents
AI Safety
Technical
I’m excited to share some findings from a recent project I undertook as part of my masters studies. I dived into the attention patterns of “sleeper agent” Large Language…
May 8, 2025
Mo Baker
No matching items
