Theme of inquiry
How do inference-time strategies and token budgets affect reasoning capability?
A question within its area, explored through 9 lines of inquiry below — each a family of specific questions the research asks.
77 specific questions
- Does task diversity in pretraining data transfer reasoning better than larger models?
- Why do reasoning gains resist clear attribution to specific training changes?
- Can smaller amounts of diverse reasoning demonstrations replace exhaustive factual training data?
- How much does pre-training frequency predict reasoning task performance?
- How do single training examples activate reasoning capabilities in language models?
- Can models maintain reasoning-output coupling while improving domain accuracy?
- Can small demonstration sets unlock general reasoning without large question data?
126 specific questions
- Can reinforcement learning add new capabilities or only remove inaccurate knowledge?
- Does RL amplify existing reasoning or create genuinely new computational strategies?
- Can RL create new reasoning primitives that pretraining never established?
- How does RL refine reasoning paths without simply adding model capability?
- Does RL primarily teach when to use reasoning or how to reason?
- Does RL refine existing knowledge or discover entirely new capabilities?
- When does RL discover genuinely novel reasoning strategies versus timing optimization?
58 specific questions
- Can parallel reasoning chains outperform longer sequential chains with the same compute?
- How do parallel and sequential retrieval strategies compare in compute efficiency?
- What makes parallel thinking more efficient than sequential chains?
- Can parallel independent reasoning outperform sequential iterative refinement?
- Can parallel retrieval chains avoid the context consumption problem?
- When does sequential chain-of-thought dramatically beat parallel voting approaches?
- Can adaptive per-step decisions outperform uniform retrieval policies across different reasoning tasks?
100 specific questions
- Why do models overthink underspecified problems instead of rejecting them?
- Why do reasoning models wander instead of searching systematically?
- Is reasoning failure caused by task complexity or training distribution gaps?
- Do reasoning models overthink ill-posed questions instead of recognizing incompleteness?
- Why do reasoning models fail on structurally unfamiliar instances?
- What mechanisms cause reasoning models to wander rather than focus?
- Why do reasoning models fail at learning hidden rules from sparse exceptions?
63 specific questions
- What makes multi-paradigm chaining a distinct reasoning topology?
- Why does augmenting symbolic reasoning outperform replacing it entirely?
- Why do format and structure matter more than actual content in reasoning?
- What distinguishes the convergence patterns between reasoning and lexical variation tasks?
- Do reasoning systems reuse cognitive structures across unrelated topics?
- Can we transfer reasoning structure without copying surface form?
- What makes structural logic correlate so strongly with contextual consistency?
64 specific questions
- Does reasoning fine-tuning actually damage a model's ability to abstain?
- Does reasoning fine-tuning actually reduce a model's ability to abstain?
- Why does reasoning fine-tuning reduce a model's ability to abstain?
- Why do instruction following and reasoning capability trade off in training?
- Can reasoning fine-tuning improve both capability and instruction compliance together?
- Does reasoning fine-tuning actually harm a model's ability to abstain?
- Why does fine-tuning degrade reasoning quality even as accuracy improves?
40 specific questions
- What causes policy entropy collapse in reasoning-focused reinforcement learning?
- Does policy entropy collapse limit how many iterations of reasoning training work?
- What happens to model reasoning when policy entropy collapses during RL?
- Why does policy entropy collapse limit reasoning and dialogue RL scaling?
- Does policy entropy collapse prevent inference-time search from finding solutions?
- Why does policy entropy collapse when scaling RL for reasoning?
- How does policy entropy during training affect search discipline during inference?
50 specific questions
- Do base models contain latent reasoning that minimal training can unlock?
- Can minimal training signals unlock latent reasoning capability in base models?
- What latent reasoning capability do base models already possess before training?
- What mechanisms activate latent reasoning capabilities already present in base models?
- Can models possess latent reasoning capability that training signals fail to unlock?
- Does latent reasoning capability exist in base models before any training?
- Can minimal training signals unlock reasoning already latent in pretrained representations?
25 specific questions
- Why do foundation models develop heuristics instead of world models?
- Why do foundation models develop task-specific heuristics instead of general world models?
- How do foundation models develop task-specific heuristics instead of world models?
- Can explicit constraint statements override the dominance of surface heuristics?
- Do foundation models develop task-specific shortcuts instead of building stable world models?
- Can surface heuristics override implicit constraints in domain-specific reasoning?
- What distinguishes task-specific heuristics from genuine world models?