Please wait while we prepare your content.
Browse unsolved research questions by category. Collaborate with the community, claim problems, and drive breakthroughs across interpretability, safety, fairness, and beyond.
Apply formal methods to prove properties of reward models used in RLHF, ensuring they preserve human intent under distribution shift.
Develop methods to identify and characterize circuits in models with >100B parameters without prohibitive computational cost.
Measure how biases in high-resource training languages transfer to low-resource language outputs during multilingual fine-tuning.
Design a classifier that detects adversarial prompt injections in real-time with <5ms added latency per request.
Achieve competitive accuracy with 10x reduction in FLOPs for on-device inference through novel quantization and pruning strategies.
Propose and validate human-aligned metrics for scoring poetry, fiction, and visual art generation beyond BLEU/FID.
Apply formal methods to prove properties of reward models used in RLHF, ensuring they preserve human intent under distribution shift.
Extend fairness metrics to handle intersectional protected groups without exponential sample size requirements.
Build a system that discovers new adversarial attack categories autonomously, beyond known jailbreak taxonomies.
Extend activation patching techniques to MoE models where expert routing introduces non-trivial causal structure.
Prove convergence bounds for asynchronous distributed training with heterogeneous hardware and variable network latency.
These problems are actively seeking collaborators. Join an existing team or bring your unique perspective to the table.
Develop methods to identify and characterize circuits in models with >100B parameters without prohibitive computational cost.
Measure how biases in high-resource training languages transfer to low-resource language outputs during multilingual fine-tuning.
Achieve competitive accuracy with 10x reduction in FLOPs for on-device inference through novel quantization and pruning strategies.
Build a system that discovers new adversarial attack categories autonomously, beyond known jailbreak taxonomies.
Have an unsolved research question? Submit it to the board. Accepted problems become visible to the entire research community and may qualify for lab funding.
Propose a Problem