Theme of inquiry
How can memory and attention systems overcome frequency and positional biases?
A question within its area, explored through 8 lines of inquiry below — each a family of specific questions the research asks.
28 specific questions
- Can fixing hallucination address AI's structural epistemic problem?
- How do external safeguards like retrieval augmentation prevent hallucination?
- Does inevitable LLM hallucination make detection metric validity critical?
- Why do language models hallucinate even with perfect training?
- Can architectural changes reduce hallucination without external retrieval or verification?
- How does interleaving reasoning with action prevent hallucination in language models?
- Why is hallucination the wrong term for all LLM false outputs?
37 specific questions
- Can mechanistic interpretability findings guide practical interventions in model design?
- Can interventions on model components prove mechanism without explaining encoding?
- How do mechanistic features compare to natural language for interpretability?
- Can mechanistic interpretability explain explanation-execution disconnection?
- How much do mechanistic interpretability findings reflect true reasoning architecture?
- What makes representation engineering better than mechanistic interpretability for detecting hidden objectives?
- How do probe-based interventions in activation space compare to mechanistic interpretability approaches?
40 specific questions
- Why does uniform memory consolidation sometimes degrade below the no-memory baseline?
- Why does LLM memory consolidation regress below no-memory baselines?
- What makes memory consolidation fragile compared to raw trajectory storage?
- Why do continuously consolidated agent memories eventually degrade below no-memory baseline?
- Why do accumulated memory systems sometimes hurt continual learning?
- Why do external memory consolidation systems fail worse than naive in-context learning on continual tasks?
- Why do accumulated memories hurt continual learning more than no memory?
22 specific questions
- Does transformer attention architecture systematically bias models toward sycophancy?
- Does transformer attention architecture inherently bias models toward sycophancy?
- What architectural features drive sycophancy closer to inference than training?
- Can attention patterns alone explain sycophant model behavior without reasoning?
- How does the U-shaped attention distribution relate to transformer sycophancy?
- Why does transformer attention architecture reinforce sycophancy and agreement?
- Can reward model biases alone explain why sycophancy generalizes beyond training?
61 specific questions
- Does moving memory outside model weights avoid the limitations of in-weight retention?
- How does in-weight memorization scale with model parameter count?
- Can in-weight memorization scale beyond model parameter count limits?
- Why does fine-tuning models for continuous reasoning cause catastrophic forgetting?
- Can models recover knowledge with completely unrelated retraining tasks?
- Why does fine-tuning for continuous space cause catastrophic forgetting?
- How does in-weights adaptation create spurious forgetting in models?
26 specific questions
- Should GUI agents use intermediate structured representations instead of raw pixels?
- Why does explicit screen parsing outperform pure vision in GUI agents?
- Can specialized perception components replace end-to-end vision in GUI agents?
- Can screen perception be effectively decoupled from planning in GUI agents?
- What role does visual perception play alongside accessibility tree information?
- Why does identifying UI element types and locations enable downstream task learning?
- Can text-based and vision-based screen understanding achieve similar performance?
45 specific questions
- How do retrieval heads achieve sparse attention naturally in transformers?
- How does transformer attention structurally bias models toward prominent and repeated content?
- Why do transformer attention patterns show positional and sequential bias across tasks?
- How does transformer attention bias toward repeated and context-prominent content?
- What role does attention structure play in creating position bias?
- Why do transformer attention mechanisms favor prominent context over factual verification?
- How do attention patterns and circuits function as algorithmic representations?
47 specific questions
- How do recurrent memory systems handle ultra-long context differently than attention?
- How do adaptive memory modules compare to feedback-based working memory for long context?
- Do long-term memory modules outperform consolidation into fast weights?
- What persistent memory architectures best support storing precomputed inferences across sessions?
- Can precomputed inferences be stored in memory modules between model interactions?
- Can compressed long-term memory outperform fixed-window token retention?
- Can adaptive memory modules combine long-term filtering with short-term attention benefits?