mlsanityData to 14 Oct 2026

Can Computation from Earlier Problems Help LLMs Solve New Ones?

arXiv 2609.39394 ↗First posted 30 Sep 2026cs.AI cs.CL cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Large language models often solve independent problems in the same conversation. Can computation from earlier problems help them solve new ones? To answer this question, we first conduct preliminary experiments showing that retained history can raise or lower later-turn accuracy, even within the same domain. To understand these effects, we use controlled replay to isolate internal state changes specific to each problem-history pairing. Across different histories, these changes preserve similar relationships among current problems. To improve reasoning under retained history, we introduce STAIR (Stale-Token Attention for Inter-query Reuse). STAIR captures keys and values from earlier response generation in a fixed bank. It learns to redirect current queries when they read this bank during prompt processing. The base model remains frozen; only 12,288 parameters are trained. Across three Qwen models and four benchmarks, STAIR improves average later-turn accuracy by up to 11.67 percentage points over the unmodified model with history.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 25 of its 35 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

25 of 35 references.

References · 25

  1. 2025Qwen3 Technical Reportcited by 8,820
  2. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  3. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  4. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,915
  5. 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,795
  6. 2023Let's Verify Step by Stepcited by 3,789
  7. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  8. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,810
  9. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,598
  10. 2023Efficient Streaming Language Models with Attention Sinkscited by 2,186
  11. 2019Transformer-XL: Attentive Language Models Beyond a Fixed-Length Contextcited by 1,850
  12. 2023GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpointscited by 1,346
13 more
  1. 2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,141
  2. 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 797
  3. 2024Gated Delta Networks: Improving Mamba2 with Delta Rulecited by 532
  4. 2025LLMs Get Lost In Multi-Turn Conversationcited by 429
  5. 2023Prompt Cache: Modular Attention Reuse for Low-Latency Inferencecited by 268
  6. 2023Think before you speak: Training Language Models With Pause Tokenscited by 264
  7. 2024CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusioncited by 223
  8. 2022Recurrent Memory Transformercited by 203
  9. 2025Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chainscited by 85
  10. 2024Deliberation in Latent Space via Differentiable Cache Augmentationcited by 26
  11. 2026Restoring Exploration after Post-Training: Latent Exploration Decoding for Large Reasoning Modelscited by 4
  12. 2026Beyond Speedup -- Utilizing KV Cache for Sampling and Reasoningcited by 3
  13. 2026Pair-In, Pair-Out: Latent Multi-Token Prediction for Efficient LLMscited by 2

Built on by

Not cited yet.