Fork-think with confidence identifies forking points via model confidence in a single path before sampling continuations, cutting tokens up to 30% and runtime up to 57% on reasoning benchmarks while matching or exceeding parallel thinking performance.
Making Language Models Better Reasoners with Step-Aware Verifier
8 Pith papers cite this work, alongside 55 external citations. Polarity classification is still indexing.
8
Pith papers citing it
55
external citations · OpenAlex
citation-role summary
method 1
citation-polarity summary
roles
method 1polarities
use method 1representative citing papers
VeryTrace formalizes CoT traces into a compilable DSL and uses a hybrid deterministic-plus-LLM verifier for step-level error detection and repair, improving zero-shot accuracy on AIME, LLM-BabyBench, and CLUTRR.
o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.
On-policy self-distillation with sampled demonstrations reduces rollout diversity by amplifying existing probability gaps in the base model, unlike ideal RL which preserves ratios among correct outputs.
Empirical analysis shows scaling inference compute via strategies like tree search can be more efficient than scaling model parameters, with 7B models plus novel search outperforming 34B models.
Math-Shepherd is an automatically trained process reward model that scores solution steps to verify and reinforce LLMs, lifting Mistral-7B from 77.9% to 89.1% on GSM8K and 28.6% to 43.5% on MATH.
An iterative bootstrapped self-filtering approach selects balanced clean and diverse subsets from noisy vision-language datasets to train improved CLIP models.
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
-
Fork-Think with Confidence
Fork-think with confidence identifies forking points via model confidence in a single path before sampling continuations, cutting tokens up to 30% and runtime up to 57% on reasoning benchmarks while matching or exceeding parallel thinking performance.
-
VeryTrace: Verifying Reasoning Traces through Compilable Formalism and Structured Verification
VeryTrace formalizes CoT traces into a compilable DSL and uses a hybrid deterministic-plus-LLM verifier for step-level error detection and repair, improving zero-shot accuracy on AIME, LLM-BabyBench, and CLUTRR.
-
Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs
o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.
-
On-Policy Self-Distillation with Sampled Demonstrations Reduces Output Diversity
On-policy self-distillation with sampled demonstrations reduces rollout diversity by amplifying existing probability gaps in the base model, unlike ideal RL which preserves ratios among correct outputs.
-
Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models
Empirical analysis shows scaling inference compute via strategies like tree search can be more efficient than scaling model parameters, with 7B models plus novel search outperforming 34B models.
-
Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations
Math-Shepherd is an automatically trained process reward model that scores solution steps to verify and reinforce LLMs, lifting Mistral-7B from 77.9% to 89.1% on GSM8K and 28.6% to 43.5% on MATH.
-
Data Selection Through Iterative Self-Filtering for Vision-Language Settings
An iterative bootstrapped self-filtering approach selects balanced clean and diverse subsets from noisy vision-language datasets to train improved CLIP models.
-
ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning
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.