Research map / Large language models / Retrieval and knowledge
Reasoning with Memory Augmented Neural Networks for Language Comprehension
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 |
|---|---|---|---|
| Use adaptive computation with a termination head instead of query-memory gating (ablation) | NSE · WDW strict test | accuracy: 65.8 -> 66.2 | Helped |
| Use adaptive computation with a termination head instead of query-memory gating (ablation) | NSE · WDW relaxed test | accuracy: 65.9 -> 66.7 | Helped |
| Increase the maximum steps from T=1 to adaptive computation T=12 (ablation) | NSE · CBT-NE test | accuracy: 71.1 -> 73.2 | Helped |
| Increase query-gating steps from T=6 to T=9 (ablation) | NSE Query Gating · CBT-NE test | accuracy: 71.4 -> 72.6 | Helped |
| Increase query-gating steps from T=9 to T=12 (ablation) | NSE Query Gating · CBT-NE test | accuracy: 72.6 -> 72.2 | Hurt |
| Increase the maximum steps from T=2 to adaptive computation T=12 (ablation) | NSE Adaptive Computation · CBT-CN test | accuracy: 71.2 -> 71.4 | Helped |
| Use adaptive computation with a termination head rather than query gating (ablation) | NSE · CBT-CN test | accuracy: 70.7 -> 71.4 | Mixed |
| Add L2 weight decay (ablation) | NSE · CBT development sets | accuracy | No clear effect |
| Increase query-gating maximum steps to T=15 (ablation) | NSE Query Gating · CBT-NE test | accuracy: 71.4 | Hurt |
Closest papers by meaning
- Phenomenal Yet Puzzling: Testing Inductive Reasoning Capabilities of Language Models with Hypothesis Refinement (ICLR 2024 · similarity 0.86)
- A Framework for Inference Inspired by Human Memory Mechanisms (ICLR 2024 · similarity 0.86)
- I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders (AAAI 2026 · similarity 0.85)
- Hypothesis Search: Inductive Reasoning with Language Models (ICLR 2024 · similarity 0.85)
- A Simple Model of Inference Scaling Laws (ICML 2025 · similarity 0.85)
- Meaningful Learning: Enhancing Abstract Reasoning in Large Language Models via Generic Fact Guidance (NeurIPS 2024 · 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.