Tagged "research"
- LLM Neuroanatomy II: Modern LLM Hacking and Hints of a Universal Language
- AI Agents Can Autonomously Perform Experimental High Energy Physics
- AI's Impact on Mathematics Analogous to Car's Impact on Cities
- Researcher Discovers Universal "Danger Zone" in Transformer Model Architecture at 50% Depth
- Memory Should Decay: Implementing Temporal Memory Decay in Local LLM Systems
- Student Researcher Achieves 42x Model Compression Through Novel Architecture
- ETH Zurich Research Challenges Context-Length Assumptions in LLM Agents
- Google Research Finds Longer Chain-of-Thought Correlates Negatively With Accuracy
- Extracting 100K Concepts from an 8B LLM