Research map / Architectures and efficiency / RNNs and normalization
Learning Memory Access Patterns
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace table-based prefetchers with an embedding LSTM | Memory prefetching · Geometric Mean | precision-at-10: 0.11 -> 0.70 | Helped |
| Replace table-based prefetchers with a clustering LSTM | Memory prefetching · Geometric Mean | precision-at-10: 0.11 -> 0.69 | Helped |
| Replace GHB with an embedding LSTM | Memory prefetching · Geometric Mean | recall-at-10: 0.39 -> 0.27 | Hurt |
| Replace the stream prefetcher with an embedding LSTM | Memory prefetching · Geometric Mean | recall-at-10: 0.72 -> 0.27 | Hurt |
| Cluster address space and model each region with a clustering LSTM | Embedding LSTM prefetching · Geometric Mean | recall-at-10: 0.27 -> 0.75 | Helped |
| Use PCs only, removing deltas from the inputs (ablation) | Embedding LSTM · Geometric Mean | precision-at-10: 0.70 -> 0.70 | No clear effect |
| Use PCs only, removing deltas from the inputs (ablation) | Embedding LSTM · Geometric Mean | recall-at-10: 0.27 -> 0.12 | Hurt |
| Use deltas only, removing PCs from the inputs (ablation) | Embedding LSTM · Geometric Mean | precision-at-10: 0.70 -> 0.61 | Hurt |
| Use deltas only, removing PCs from the inputs (ablation) | Embedding LSTM · Geometric Mean | recall-at-10: 0.27 -> 0.24 | No clear effect |
Closest papers by meaning
- Memory Layers at Scale (ICML 2025 · similarity 0.88)
- Adam-mini: Use Fewer Learning Rates To Gain More (ICLR 2025 · similarity 0.86)
- Recurrent neural networks: vanishing and exploding gradients are not the end of the story (NeurIPS 2024 · similarity 0.85)
- Reinforced Fast Weights via Next-Sequence Prediction (NeurIPS 2026 · similarity 0.85)
- The Geometric Inductive Bias of Grokking: Bypassing Phase Transitions via Architectural Topology (NeurIPS 2026 · similarity 0.85)
- Accelerating Linear Recurrent Neural Networks for the Edge with Unstructured Sparsity (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.