Research map / Large language models / Instruction and fine-tuning
Predicting Concrete and Abstract Entities in Modern Poetry
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 |
|---|---|---|---|
| fine-tune the Wikipedia-pretrained BiLSTM on modern American poetry | BiLSTM language model · Exp I, synonyms, random test split | MRR: 0.435 -> 0.627 | Helped |
| fine-tune on poetry, evaluated on unseen authors | BiLSTM language model · Exp I, synonyms, unseen authors | MRR: 0.433 -> 0.513 | Helped |
| fine-tune on poetry, same-category candidates, random split | BiLSTM language model · Exp II, same-category candidates, random test | MRR: 0.397 -> 0.518 | Helped |
| fine-tune on poetry, same-category candidates, unseen authors | BiLSTM language model · Exp II, unseen authors | MRR: 0.424 -> 0.426 | No clear effect |
| fine-tune on poetry, 4 candidates | BiLSTM language model · Exp III, 75 held-out verses | MRR: 0.535 -> 0.640 | Helped |
| fine-tune on poetry, 4 candidates | BiLSTM language model · Exp III, 75 held-out verses | accuracy (%): 26.7 -> 40.0 | Helped |
| restrict candidates from synonyms to hypernym-hyponym related words, unseen authors (ablation) | fine-tuned BiLSTM · unseen authors | MRR: 0.513 -> 0.426 | Hurt |
| evaluate on unseen authors instead of random test sentences (ablation) | fine-tuned BiLSTM · Exp I, synonyms | MRR: 0.627 -> 0.513 | Hurt |
| fine-tuned model vs professional poets | fine-tuned BiLSTM · Exp III | accuracy (%): 38.7 -> 40.0 | Helped |
| split improvement by concrete vs abstract entities | fine-tuned BiLSTM · Exp I | error reduction | Mixed |
Closest papers by meaning
- Metaphor Is Not All Attention Needs (NeurIPS 2026 · similarity 0.83)
- PoetryDiffusion: Towards Joint Semantic and Metrical Manipulation in Poetry Generation (AAAI 2024 · similarity 0.83)
- Seeking the Unfamiliar but Memorable: Conceptual Creativity as Meta-Learning (NeurIPS 2026 · similarity 0.81)
- The Emergence of Abstract Thought in Large Language Models Beyond Any Language (NeurIPS 2025 · similarity 0.81)
- AI as Humanity’s Salieri: Quantifying Linguistic Creativity of Language Models via Systematic Attribution of Machine Text against Web Text (ICLR 2025 · similarity 0.81)
- Roll the dice & look before you leap: Going beyond the creative limits of next-token prediction (ICML 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.