Research map / Large language models / Language, safety and interpretability
Multi-Task Learning For Parsing The Alexa Meaning Representation Language
Vittorio Perera, Tagyoung Chung, Thomas Kollar, Emma Strubell
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 SLU slot embeddings (serial pipe) as extra task (ablation) | baseline bi-LSTM multi-task AMRL parser · AMRL test | IRER: 25.7176 -> 23.1312 | Improved |
| add SLU domain embeddings on top of slot embeddings (ablation) | baseline bi-LSTM multi-task AMRL parser · AMRL test | IRER: 23.1312 -> 23.0222 | Improved |
| serial parallel vs pipe slot embeddings: use pipe instead of parallel (ablation) | +SLU Slots · AMRL test | IRER: 24.2876 -> 23.1312 | Improved |
| use a single shared slot/domain embedding layer feeding the type layer (pipe) (ablation) | +SLU Slots/Domain · AMRL test | IRER: 23.0222 -> 29.9418 | Hurt |
| joint training of SLU and AMRL instead of pretraining (ablation) | SLU-augmented models · AMRL test | IRER: 23.1312 -> 25.3534 (Slots pipe); 29.9418 -> 24.4501 (Slots/Domain pipe) | Mixed |
| add custom constrained beam decoder (IOB and ontology constraints) (ablation) | proposed SLU model · AMRL test | IRER: decreases IRER by 1% absolute | Improved |
| transfer SLU representations vs adding gazetteers and word2vec | baseline · AMRL test | IRER: decreases full-parse error by 1.5% IRER | Improved |
| add intent embeddings from SLU (ablation) | initial models · AMRL test | accuracy | No clear effect |
Closest papers by meaning
- An Information-theoretic Multi-task Representation Learning Framework for Natural Language Understanding (AAAI 2025 · similarity 0.85)
- MMRL: Multi-Modal Representation Learning for Vision-Language Models (CVPR 2025 · similarity 0.84)
- MMSU: A Massive Multi-task Spoken Language Understanding and Reasoning Benchmark (ICLR 2026 · similarity 0.84)
- MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark (NeurIPS 2024 · similarity 0.84)
- EmbedLLM: Learning Compact Representations of Large Language Models (ICLR 2025 · similarity 0.84)
- Banyan: Improved Representation Learning with Explicit Structure (ICML 2025 · similarity 0.84)
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.
