When Can LLM Reasoning Reliably Reuse Intermediate Results?
Sampling complete solutions gives large language models more chances to solve a problem, but each attempt starts again.
ProofPaper ↗
Key points
- For tasks with several subproblems, all required answers must be correct in the same attempt.
- Retaining intermediate results can avoid repeating correct work, but errors in those results may affect later attempts.
- Experiments across eight reasoning models examine composite mathematics and arithmetic.
- Across four DeepSeek models, average success on the second subproblem is 36.2 percent without first-subproblem context, 57.2 percent with its answer, and 58.3 percent with its reasoning and answer.
Sources (1)
- [1]When Can LLM Reasoning Reliably Reuse Intermediate Results?arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 05:24 AM
Sampling complete solutions gives large language models more chances to solve a problem, but each attempt starts again.
For tasks with several subproblems, all required answers must be correct in the same attempt.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 20, 2026Do Diffusion LLMs Scale? Testing Mercury 3 + Decide
- Oct 19, 2026Introducing Clairic, a new System One class of extraction models
- Oct 19, 2026Beyond Attack Success Rate: Cost-Aware Metric for Evaluating Iterative LLM Jailbreaks
- Oct 19, 2026MoveBench: A Rubric-Based Benchmark for Evaluating Large Language Models in Sports and Health Coaching
- Oct 19, 2026Learning Who Talks to Whom: Role-Aware Communication Topology Scheduling for Efficient LLM Multi-Agent Collaboration
- Oct 18, 2026Elevatormusic/DeepSeek-V4.1-Flash-1017-EXL3-TensorFold-2x-DGX-Spark: DeepSeek-V4.1-Flash-1017 EXL3 2.9bpw on 2x NVIDIA DGX Spark (GB10) with TensorFold: 2.1-2.3x the vLLM kit's decode speed.