DEV Community

#interpretability

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
J-space in practice: using Anthropic's Jacobian lens to decide what an LLM can forget
anyesh profile

J-space in practice: using Anthropic's Jacobian lens to decide what an LLM can forget

#jspace #jacobianlens #interpretability #kvcache
Comments
6 min read
Beyond Reconstruction: Verifying Model Explanations with RECAP
pneumetron profile

Beyond Reconstruction: Verifying Model Explanations with RECAP

#interpretability #mechanisticinterpretability #aisafety #recap
Comments
3 min read
Length Penalties in LLMs: Shorter Chains of Thought, Hidden Influences
pneumetron profile

Length Penalties in LLMs: Shorter Chains of Thought, Hidden Influences

#llm #chainofthought #reinforcementlearning #interpretability
Comments
3 min read
The safety switch that doesn't actually work
breachprotocol profile

The safety switch that doesn't actually work

#interpretability #safety #sparseautoencoders
Comments
4 min read
Claude Was Always Thinking Ahead. Now We Can Read It.
michaeltuszynski profile

Claude Was Always Thinking Ahead. Now We Can Read It.

#interpretability #airesearch #claudeai #anthropic
2
Comments
7 min read
Mechanistic Interpretability is a 2026 Breakthrough Technology. Here's What That Means for the "LLMs Are Just Matrix Multiplication" Debate
ikramar profile

Mechanistic Interpretability is a 2026 Breakthrough Technology. Here's What That Means for the "LLMs Are Just Matrix Multiplication" Debate

#discuss #ai #machinelearning #interpretability
3
Comments
10 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.