Research map / Architectures and efficiency / RNNs and normalization
SkipFlow: Incorporating Neural Coherence Features for End-to-End Automatic Text Scoring
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 |
|---|---|---|---|
| Add tensor-based SkipFlow coherence features to a mean-pooled LSTM | LSTM (d=50) · ASAP | QWK: 0.646 -> 0.764 | Helped |
| Use tensor composition instead of bilinear composition in SkipFlow (ablation) | SkipFlow LSTM · ASAP | QWK: 0.753 -> 0.764 | Helped |
| Add SkipFlow coherence features to CNN representations (ablation) | CNN · ASAP | QWK: 0.708 -> 0.696 | Hurt |
| Use mean pooling instead of only the last recurrent output (ablation) | RNN, GRU, and LSTM · ASAP | QWK: RNN: 0.102 -> 0.645; GRU: 0.418 -> 0.646; LSTM: 0.467 -> 0.646 | Helped |
| Increase the number of tensor slices from the small or large extremes toward 4 to 6 (ablation) | SkipFlow LSTM (Tensor) · ASAP prompts | QWK | Mixed |
| Increase relevance width from a small width toward 50 to 100 (ablation) | SkipFlow LSTM (Tensor) · ASAP prompts | QWK | Helped |
| Increase LSTM dimensionality from d=50 to d=300 (ablation) | LSTM · ASAP | QWK: 0.646 -> 0.715 | Helped |
| Use a bidirectional LSTM instead of a unidirectional mean-pooled LSTM (ablation) | LSTM · ASAP | QWK: 0.646 -> 0.667 | No clear effect |
Closest papers by meaning
- Learning to Extract Coherent Summary via Deep Reinforcement Learning (AAAI 2018 · similarity 0.84)
- BBScore: A Brownian Bridge Based Metric for Assessing Text Coherence (AAAI 2024 · similarity 0.84)
- Automatically Interpreting Millions of Features in Large Language Models (ICML 2025 · similarity 0.82)
- Learning the Signature of Memorization in Autoregressive Language Models (NeurIPS 2026 · similarity 0.81)
- Understanding the Differences in Foundation Models: Attention, State Space Models, and Recurrent Neural Networks (NeurIPS 2024 · similarity 0.81)
- Extended LSTMs for Knowledge Tracing: Peeking Inside the Black Box (Student Abstract) (AAAI 2025 · similarity 0.81)
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.