Research map / Architectures and efficiency / RNNs and normalization
Non-autoregressive translation
Speeds machine translation decoding with non-autoregressive models and efficient decoder designs.
102 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in RNNs and normalization, keyword cluster “end · differentiable · translation”
State History Benchmarks Tried Baselines Preprints Who Export
Where it stands
Quiet None of the tracked benchmarks has a report from 2025–26.
10 papers in 2023–24 and 7 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 1.5% to 0.7%.
No longer reported by papers here: WMT14 De-En (BLEU, last in 2023); WMT14 En-De (BLEU, last in 2023); IWSLT14 De-En (BLEU, last in 2023); WMT16 En-Ro (BLEU, last in 2023); WMT'16 RO-EN (BLEU, last in 2023).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Deep Speech 2 : End-to-End Speech Recognition in English and Mandarin (ICML 2016 · 2,239 citations)
- 2017 · Machine Comprehension Using Match-LSTM and Answer Pointer (ICLR 2017 · 456 citations)
- 2018 · Non-Autoregressive Neural Machine Translation (ICLR 2018 · 507 citations)
- 2019 · Non-Autoregressive Machine Translation with Auxiliary Regularization (AAAI 2019 · 148 citations)
- 2020 · Neural Machine Translation with Byte-Level Subwords (AAAI 2020 · 137 citations)
- 2021 · Guiding Non-Autoregressive Neural Machine Translation Decoding with Reordering Information (AAAI 2021 · 68 citations)
- 2022 · MLSLT: Towards Multilingual Sign Language Translation (CVPR 2022 · 49 citations)
- 2023 · Prompting Neural Machine Translation with Translation Memories (AAAI 2023 · 18 citations)
- 2024 · FLAME: A Small Language Model for Spreadsheet Formulas (AAAI 2024 · 5 citations)
- 2025 · Geo-Sign: Hyperbolic Contrastive Regularisation for Geometrically Aware Sign Language Translation (NeurIPS 2025 · 3 citations)
- 2026 · MineDraft: A Framework for Batch Parallel Speculative Decoding (ICML 2026 · 0 citations)
Earliest papers here
- Deep Speech 2 : End-to-End Speech Recognition in English and Mandarin (ICML 2016 · 2,239 citations)
- Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models (AAAI 2016 · 1,724 citations)
- Syntactic Skeleton-Based Translation (AAAI 2016 · 10 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
WMT14 De-En · BLEU (higher is better)
9 papers, 2018–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 31.9 | Model-Level Dual Learning (ICML 2018 · 33 citations) |
WMT14 En-De · BLEU (higher is better)
9 papers, 2018–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 28.9 | Model-Level Dual Learning (ICML 2018 · 33 citations) |
IWSLT14 De-En · BLEU (higher is better)
8 papers, 2018–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 35.19 | Model-Level Dual Learning (ICML 2018 · 33 citations) |
| 2020 | 37 | Data Diversification: A Simple Strategy For Neural Machine Translation (NeurIPS 2020 · 43 citations) |
WMT16 En-Ro · BLEU (higher is better)
6 papers, 2019–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 34.51 | Non-Autoregressive Neural Machine Translation with Enhanced Decoder Input (AAAI 2019 · 119 citations) |
WMT'16 RO-EN · BLEU (higher is better)
4 papers, 2021–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 31.99 | Guiding Non-Autoregressive Neural Machine Translation Decoding with Reordering Information (AAAI 2021 · 68 citations) |
| 2022 | 34.6 | Non-autoregressive Translation with Layer-Wise Prediction and Deep Supervision (AAAI 2022 · 25 citations) |
What has been tried
93 papers here report 1,017 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 8%, hurt 20%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Reduce model size from 60M to 16M parameters | Formula completion, 0.90 prefix · exact match | 0.70 -> 0.59 | Hurt | FLAME: A Small Language Model for Spreadsheet Formulas (AAAI 2024 · 5 citations) |
| Use token-level instead of pooled global alignment | CSL-Daily test · BLEU-4 | 27.17 -> 27.42 | Helped | Geo-Sign: Hyperbolic Contrastive Regularisation for Geometrically Aware Sign Language Translation (NeurIPS 2025 · 3 citations) |
| Exclude duration 0 | English leaderboard datasets (AVG) · WER (%) | 7.10 -> 6.98 | Helped | HAINAN: Fast and Accurate Transducer for Hybrid-Autoregressive ASR (ICLR 2025 · 0 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| Transformer | IWSLT14 De->En | BLEU | 32.55 – 34.7 | 3 papers, e.g. AAAI 2019 and NeurIPS 2020 |
| Vanilla NAT | WMT'14 EN-DE | BLEU | 19.34 – 21.18 | 2 papers, e.g. AAAI 2023 and AAAI 2022 |
| Transformer | WMT16 EN→RO | BLEU | 32.85 – 35.61 | 2 papers, e.g. AAAI 2020 and AAAI 2019 |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 17 papers; from OpenAlex and the papers' affiliations).
- Chinese Academy of Sciences · 2 papers
- Georgia Institute of Technology · 2 papers
Use it
Download BibTeX of 30 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Differentiable optimization layers (125 papers)
- Neural-symbolic reasoning (98 papers)
- Learned embeddings for retrieval (76 papers)
- Associative memory networks (48 papers)
Elsewhere in the topic
- Efficient network compression and training (229 papers)
- Theory of depth and width (223 papers)
- Representation geometry and collapse (169 papers)
- Dataset distillation and data-efficient training (158 papers)
- Neural scaling laws (155 papers)
- Distributed LLM inference systems (147 papers)
- Layer-wise analysis of language models (145 papers)
- Dataset condensation and flow estimation (126 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).