Research map / Architectures and efficiency / RNNs and normalization
Decoding with Value Networks for Neural Machine Translation
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 value network reranking during beam search (NMT-VNN) vs classic beam search | RNN NMT · En→Fr newstest2014 | BLEU: +1.03 | Helped |
| add value network reranking during beam search (NMT-VNN) vs classic beam search | RNN NMT · En→De newstest2014 | BLEU: +1.3 | Helped |
| value network decoding instead of beam search optimization (NMT-BSO) | RNN NMT · En→Fr newstest2014 | BLEU: +0.31 | Helped |
| value network decoding instead of beam search optimization (NMT-BSO) | RNN NMT · En→De newstest2014 | BLEU: +0.33 | Helped |
| add value network reranking during beam search vs classic beam search | RNN NMT · Zh→En NIST06 | BLEU: +1.4 | Helped |
| add value network reranking during beam search vs classic beam search | RNN NMT · Zh→En NIST08 | BLEU: +1.82 | Helped |
| value network decoding instead of beam search optimization (NMT-BSO) | RNN NMT · Zh→En NIST06 | BLEU: +1.01 | Helped |
| value network decoding instead of beam search optimization (NMT-BSO) | RNN NMT · Zh→En NIST08 | BLEU: +0.72 | Helped |
| add value network decoding to a deep 4-layer LSTM NMT model | deep 4-layer LSTM NMT · En→Fr | BLEU: +0.33 | Helped |
| increase beam size with value network (NMT-VNN) vs without (ablation) | RNN NMT · En→De beam size sweep | BLEU | Helped |
| value network decoding hyperparameter alpha varied 0.7 to 0.95 (ablation) | NMT-VNN · En→De | BLEU | No clear effect |
Closest papers by meaning
- Empirical Analysis of Beam Search Performance Degradation in Neural Sequence Models (ICML 2019 · similarity 0.87)
- Dual Learning for Machine Translation (NeurIPS 2016 · similarity 0.87)
- From Fully Trained to Fully Random Embeddings: Improving Neural Machine Translation with Compact Word Embedding Tables (AAAI 2022 · similarity 0.87)
- Analyzing Uncertainty in Neural Machine Translation (ICML 2018 · similarity 0.87)
- Synchronous Interactive Decoding for Multilingual Neural Machine Translation (AAAI 2021 · similarity 0.87)
- Calibrating Translation Decoding with Quality Estimation on LLMs (NeurIPS 2025 · similarity 0.86)
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.