REVIEW 11 cited by
Efficient Test-Time Scaling via Self-Calibration
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Increasing test-time computation is a straightforward approach to enhancing the quality of responses in Large Language Models (LLMs). While Best-of-N sampling and Self-Consistency with majority voting are simple and effective, they require a fixed number of sampling responses for each query, regardless of its complexity. This could result in wasted computation for simpler questions and insufficient exploration for more challenging ones. In this work, we argue that model confidence of responses can be used for improving the efficiency of test-time scaling. Unfortunately, LLMs are known to be overconfident and provide unreliable confidence estimation. To address this limitation, we introduce Self-Calibration by distilling Self-Consistency-derived confidence into the model itself. This enables reliable confidence estimation at test time with one forward pass. We then design confidence-based efficient test-time scaling methods to handle queries of various difficulty, such as Early-Stopping for Best-of-N and Self-Consistency with calibrated confidence. Experiments on three LLMs across six datasets demonstrate the effectiveness of our approach. Specifically, applying confidence-based Early Stopping to Best-of-N improves MathQA accuracy from 81.0 to 83.6 with a sample budget of 16 responses, indicating the efficacy of confidence-based sampling strategy at inference time.
Forward citations
Cited by 11 Pith papers
-
Parallel-R1: Towards Parallel Thinking via Reinforcement Learning
Parallel-R1 uses SFT cold-start on easy math plus GRPO on hard math to instill parallel thinking in Qwen3-4B, reporting 8.4% average accuracy gains and a 42.9% AIME25 gain from a parallel-exploration scaffold.
-
AALC: Large Language Model Efficient Reasoning via Adaptive Accuracy-Length Control
AALC, an accuracy-aware length reward, reduces reasoning-model response length by over 50% on math benchmarks while preserving or improving accuracy.
-
How Far Are We from Optimal Reasoning Efficiency?
The authors define a reasoning efficiency frontier and a gap metric (REG), then train models with REO-RL to shrink the gap by at least 50% with only small accuracy losses.
-
AdaDecode: Accelerating LLM Decoding with Adaptive Layer Parallelism
AdaDecode speeds up LLM generation by predicting tokens at early layers when confidence is high, running the skipped layers in parallel, and verifying the output exactly matches standard decoding.
-
POSS: Position Specialist Generates Better Draft for Speculative Decoding
Using position-specialized draft layers instead of one single draft model improves later-token acceptance in speculative decoding, yielding modest speedups on Llama-3-8B and Llama-2-13B.
-
Generalizable LLM Learning of Graph Synthetic Data with Post-training Alignment
Post-training alignment on synthetic graph data yields statistically significant gains on some real-world implicit-graph tasks, but gains are inconsistent and multi-step reasoning remains fragile.
-
CGES: Confidence-Guided Early Stopping for Efficient and Accurate Self-Consistency
Confidence-Guided Early Stopping stops querying an LLM once one candidate answer accumulates enough Bayesian posterior mass, cutting average calls from 16 to 4.9 on five reasoning benchmarks with negligible average ac...
-
Confidence-Weighted Token Set Cover for Early Hypothesis Pruning in Self-Consistency
A weighted set cover algorithm prunes intermediate hypotheses during self-consistency, cutting token use by 10 to 35 percent across five LLMs on three math benchmarks while keeping accuracy.
-
Reasoning on a Budget: A Survey of Adaptive and Controllable Test-Time Compute in LLMs
The survey's L1/L2 taxonomy and benchmark show that current reasoning models waste compute on easy problems and underthink hard ones, motivating more adaptive inference.
-
Revisiting Test-Time Scaling: A Survey and a Diversity-Aware Method for Efficient Reasoning
ADAPT, a diversity-aware prefix fine-tuning method, improves best-of-N sampling efficiency for a 1.5B reasoning model, reaching 80% accuracy at N=32 versus N=256 for the baseline.
-
Interpretable Adaptive Sampling for LLM Test-Time Scaling
A fuzzy controller that allocates a per-prompt sampling budget keeps LLM accuracy near a fixed full-budget baseline while reducing the average number of candidate answers on some datasets.
Discussion (0). Sign in to comment.