DASH assigns segment-level credit in reasoning traces using drift toward ground-truth answers, yielding 50.8% accuracy on AIME25 versus 45.4% for GRPO while reducing overthinking behaviors.
hub
S1: Simple test-time scaling
18 Pith papers cite this work. Polarity classification is still indexing.
18
Pith papers citing it
hub tools
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
ATLAS introduces an LLM-orchestrated agentic framework for dynamic test-time scaling via extensible 'explore' actions, achieving higher accuracy with fewer API calls than fixed-workflow baselines on four benchmarks.
LaneRoPE adds an inter-sequence attention mask and extended RoPE to enable collaborative parallel sequence generation in LLMs, yielding accuracy gains on math reasoning under length limits.
Document-level machine translation followed by segment-level LLM refinement provides the strongest and most stable improvements in literary translation quality, mainly enhancing fluency and style rather than adequacy.
Abstract-CoT lets models reason with short discrete latent token sequences from a reserved vocabulary, using warm-up training and RL to match verbal CoT performance with up to 11.6x fewer tokens.
ReasoningBank distills generalizable reasoning strategies from agent successes and failures to enable self-evolution, with memory-aware test-time scaling amplifying gains over raw-trajectory or success-only memory on web and software benchmarks.
SpenseGPT introduces a hybrid sparse-dense weight format and one-shot pruning that delivers 1.2x end-to-end LLM decoding speedup on B200 GPUs with FP8 while preserving accuracy on Qwen3-32B and Seed-OSS-36B.
Bucket-Level MOO reformulates multilingual fine-tuning as localized multi-objective optimization and proves it enforces a tighter Pareto stationarity condition while improving cross-lingual performance on four LLMs.
Base LLMs show latent judge calibration that Self-Evaluation Elicitation (SEE) surfaces with 160 examples via RL calibration followed by masked distillation.
RISC reformulates self-consistency answer selection as a ranking task solved by a lightweight LambdaRank model with five hand-designed features, yielding better accuracy-efficiency trade-offs than majority voting on QA benchmarks.
Decoding Time Verification (DTV) interleaves verifier calls at structural boundaries during autoregressive code generation for C-to-Rust and JavaScript-to-TypeScript translation, raising pass rates while using fewer tokens than post-hoc baselines.
Prefix consistency weights CoT answers by their regeneration frequency from truncated prefixes and reaches standard self-consistency accuracy at a median 4.6x fewer tokens across five models and four benchmarks.
RLCM trains LLMs with a margin-enhanced process reward that widens the gap between correct and incorrect reasoning steps, improving calibration on math, code, logic, and science tasks without hurting accuracy.
SimpleTES scales test-time evaluation in LLMs to discover state-of-the-art solutions on 21 scientific problems across six domains, outperforming frontier models and optimization pipelines with examples like 2x faster LASSO and new Erdos constructions.
AI model failures on complex tasks become increasingly incoherent with longer reasoning chains, making consistent misalignment less likely than chaotic errors as capabilities scale.
CAT uses intrinsic confidence signals in preference optimization to adapt reasoning length in LRMs, outperforming uniform compression baselines on accuracy across benchmarks.
Learned multi-feature stopping improves accuracy-cost tradeoffs on free-form math but scalar rules match or exceed it on multiple-choice and hard problems.
ThinkBooster supplies a modular library, joint performance-efficiency benchmark, and deployable proxy for test-time compute scaling of LLM reasoning on math and coding tasks.
citing papers explorer
-
LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation
LaneRoPE adds an inter-sequence attention mask and extended RoPE to enable collaborative parallel sequence generation in LLMs, yielding accuracy gains on math reasoning under length limits.
-
ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
ReasoningBank distills generalizable reasoning strategies from agent successes and failures to enable self-evolution, with memory-aware test-time scaling amplifying gains over raw-trajectory or success-only memory on web and software benchmarks.
-
Multilingual Fine-Tuning via Localized Gradient Conflict Resolution
Bucket-Level MOO reformulates multilingual fine-tuning as localized multi-objective optimization and proves it enforces a tighter Pareto stationarity condition while improving cross-lingual performance on four LLMs.
-
The Hot Mess of AI: How Does Misalignment Scale With Model Intelligence and Task Complexity?
AI model failures on complex tasks become increasingly incoherent with longer reasoning chains, making consistent misalignment less likely than chaotic errors as capabilities scale.
-
When Does Learning to Stop Help? A Cost-Aware Study of Early Exits in Reasoning Models
Learned multi-feature stopping improves accuracy-cost tradeoffs on free-form math but scalar rules match or exceed it on multiple-choice and hard problems.