Papers All the Way Down

This is an incomplete, probably not regularly updated, list of papers I have read or plan to read.

Papers to Read

---

Safety:

• Towards Monosemanticity: Decomposing Language Models With Dictionary Learning
• Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
• Circuits
• Simple probes can catch sleeper agents

Transformers:

• RingAttention
• DiT

Image generation:

• DCGAN
• WGAN
• TryOnDiffusion

Reinforcement learning:

• TRPO
• PPO
• DPO
• AlphaFold

Miscellaneous:

• PINN
• RL and PGM

Notable Papers I’ve read

---

Safety, Alignment, and Other Things

• RLHF
• Sleeper agents
• Scaling
• Are Emergent Abilities of Large Language Models a Mirage?

Transformers

• The original attention paper!
• An Image is Worth 16x16 Words

Diffusion

• Deep Unsupervised Learning using Nonequilibrium Thermodynamics
• DDPM
• Lilan Weng’s Great Blog on Diffusion Models
  • The Annotated Diffusion Model