Pith. sign in

REVIEW 5 major objections 5 minor 3 cited by

Large Language Models Are Read/Write Policy-Makers for Simultaneous Generation

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that an off-the-shelf large language model can act as both the policy-maker and the generator in simultaneous generation, deciding on its own when to read more input and when to write output, and that this beats…

desk verdict A novel, useful simultaneous-generation recipe whose headline claim is undercut by its own ablation: the developer-set range window, not the LLM, is doing the policy work. read the letter →

arxiv 2501.00868 v1 pith:LIZ2WLZV submitted 2025-01-01 cs.CL

Shoutao Guo , Shaolei Zhang , Zhengrui Ma , Yang Feng This is my paper · ORCID
classification cs.CL
keywords simultaneousgenerationlargelanguagemodelsread/writepolicymachinetranslationstreamingautomaticspeechrecognitionwait-kKLdivergencelatency-qualitytrade-off
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to remove the need for complex dynamic-programming policy training in simultaneous generation. It proposes LSG, in which an off-the-shelf large language model both decides when to read more input and when to write output, using the wait-1 policy as a minimum-latency baseline. At each step the LLM compares its next-token distribution on the current streaming prefix with its distribution on the wait-1 prefix; when the KL divergence between these exceeds a threshold $\delta$, it writes, otherwise it reads. A confidence condition catches cases where the baseline already provides enough information, and a range constraint keeps latency inside a developer-set window. The paper reports state-of-the-art quality-latency trade-offs on simultaneous text translation, simultaneous speech translation, and streaming ASR using open-source LLMs.

What carries the argument

The mechanism is a two-distribution comparison made by the same LLM at each step. LSG computes the next-token distribution from the currently available source, $p(y_i|x_{\le j}, y_{<i})$, and from the wait-1 baseline input, $p(y_i|x_{\le i}, y_{<i})$, and takes the READ action unless $D_{\mathrm{KL}}(p(y_i|x_{\le j}, y_{<i}) \,\|\, p(y_i|x_{\le i}, y_{<i})) > \delta$. The confidence condition $\max p(y_i|x_{\le j}, y_{<i}) > \alpha$ handles false negatives where the baseline already supplies enough source information, and the $[L,U]$ range constraint caps how far the policy can stray from a fixed latency window. The wait-1 policy is the chosen baseline because it is the smallest usable input prefix, making it the natural anchor against which extra reading is measured.

What would settle it

Run LSG on a language pair with heavy reordering (for example Japanese\textrightarrow English) and compare its read/write decisions against oracle aligned-source decisions; if the KL threshold frequently disagrees with the oracle and the confidence condition is what actually triggers writes, the central policy signal is not doing the work. Alternatively, replace the two-distribution comparison with a fixed wait-k schedule using the same $[L,U]$ range constraint and the same LLM; if the fixed schedule matches LSG's quality at every latency, the distribution signal adds nothing.

Watch

Extended reading notes

Core claim

The central discovery is that policy-making for streaming generation can be derived from the LLM's own next-token distributions instead of being learned through training. Given the source prefix $x_{\le j}$ and previously generated words $y_{<i}$, the LLM produces $p(y_i|x_{\le j}, y_{<i})$, and LSG writes exactly when $D_{\mathrm{KL}}(p(y_i|x_{\le j}, y_{<i}) \,\|\, p(y_i|x_{\le i}, y_{<i})) > \delta$. The authors show that this distribution difference grows as more source content arrives and becomes significant once the necessary content is available, while a small remaining difference means the wait-1 prefix is sufficient. The confidence condition $\max p(y_i|x_{\le j}, y_{<i}) > \alpha$ allows a write even when the KL condition is not met, and the range constraint $g_i \in [\min(L+i-1,J), \min(L+i-1+U,J)]$ keeps the policy within controllable latency. The claim is that this combination of two probability tests and one window constraint lets one off-the-shelf LLM act as both policy-maker and generator, achieving state-of-the-art results across three streaming tasks.

Load-bearing premise

The load-bearing premise is that the KL divergence between the LLM's next-token distribution on the current prefix and its distribution on the wait-1 prefix is a reliable, monotone signal that enough source content has been read, so that crossing a fixed threshold $\delta$ marks the right moment to write.

Editorial extensions

If this is right

  • No dynamic-programming policy learning is needed; the policy is derived from the LLM's own distributions at inference time, and the only training used in the text experiments is a standard LoRA fine-tuning of the generator.
  • The same LSG recipe transfers across modalities, since it works with Llama2-7B-chat on text-to-text translation and with Qwen-Audio on speech-to-text translation and streaming ASR.
  • Latency and quality are controlled by the threshold $\delta$, the confidence $\alpha$, and the range window $[L,U]$, so an engineer can set a worst-case response delay in advance.
  • Better LLMs produce better simultaneous generation under LSG, as Bayling-13B outperforms Bayling-7B on Chinese\textrightarrow English at the same latency levels.
  • LSG's policies have higher sufficiency than wait-k and HMT policies applied to the same LLM, meaning the model writes more often only after the aligned source information is available.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the KL divergence between current-prefix and wait-1-prefix distributions is a general signal of source sufficiency, the same two-condition recipe should transfer to other streaming text tasks such as incremental summarization or live code completion, where an explicit alignment signal is unavailable.
  • The paper's ablation shows the range constraint is the largest single contributor, so the marginal benefit of the KL signal over a fixed wait-k schedule confined to the same $[L,U]$ window is an open question that the paper does not isolate.
  • A direct test of the policy signal would be to run LSG on a language pair with extreme word-order divergence and check whether the $\delta$ threshold, rather than the $\alpha$ confidence rule, triggers the writes that match oracle aligned-source decisions.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes LSG, a framework in which an off-the-shelf LLM serves as both generator and read/write policy-maker for simultaneous generation. At each step the LLM computes the KL divergence between its next-token distribution given the current source prefix and given the wait-1 baseline prefix; if this divergence exceeds a threshold δ, or if the top probability exceeds α, the LLM writes a token, and the allowed write positions are constrained to a developer-set window [L+i−1, L+i−1+U]. Experiments on simultaneous text translation (WMT15 De→En, MuST-C En→De), simultaneous speech translation, and streaming ASR on CoVoST2 Fr→En claim state-of-the-art latency-quality trade-offs.

Significance. If the claimed behavior holds, LSG would provide a training-free alternative to DP-trained policies for simultaneous generation with LLMs, and the paper's breadth across three tasks is valuable. The paper ships code, includes ablations and a sufficiency-rate analysis, and uses open-source models, which aids reproducibility. However, the specific evidence that the LLM's distribution-comparison signal is doing the policy work is weak; the range constraint appears to carry most of the policy burden, and key operating points are chosen per dataset without sensitivity analysis.

major comments (5)
  1. [Ablation Study, Table 2] The w/o Range ablation is the most direct evidence against the central claim of LLM-driven policy-making. Without Eq. 5's window, the method yields BLEU 21.95 at AL 3.62, nearly 10 BLEU below the full LSG at AL 4.42, and only reaches BLEU 29.90 even at AL 12.91, which is far worse than the full method at AL 7.37. This indicates that the KL/confidence conditions in Eqs. 3-4 do not by themselves identify reliable write positions; the hand-set window supplies the actual policy. To support the claim that the LLM decides timing, the paper should report, within the window, how often and how far the realized write positions depart from a trivial rule (e.g., always write at the lower bound or upper bound), and compare against a random-in-window baseline.
  2. [System Settings and Appendix Tables 5-7] δ and α are set per dataset/direction (e.g., δ=9.0, α=0.6 for De→En; δ=7.0, α=0.5 for Fr→En) with no sensitivity analysis. Because the policy is an empirical heuristic, the reported latency-quality curves may reflect tuned operating points. Please provide results over a grid of δ and α at a fixed [L,U], or show that the method is robust to these thresholds, so the reader can assess whether the comparisons are fair.
  3. [Main Results, Table 1] The streaming ASR table lists two unlabeled LSG rows (AL 3161.25, WER 31.71; AL 4342.23, WER 23.76) without stating which [L,U] and thresholds they correspond to. The baselines are reported at a single AL of 5684.38, so the claim of 'comparable quality with a delay of about 3 seconds' is supported by only one point and the higher-quality LSG point is at a latency below the baselines' latency. Please provide a full latency-WER curve with labeled operating points and have the baselines evaluated at multiple latencies where possible.
  4. [Range constraint, Eq. 5] The enforcement mechanism of the range constraint is not specified. It is unclear whether the system is forced to write when j reaches the upper bound L+i−1+U, or whether the constraint only limits when the KL/confidence conditions are consulted. This is load-bearing because the policy's behavior at the boundary determines the realized latency and quality. Please state the algorithm explicitly (e.g., pseudocode) for a single step, including what happens if neither Eq. 3 nor Eq. 4 is satisfied at the upper bound.
  5. [Introduction, Figure 1] The motivating claim that the KL divergence between current and wait-1 distributions is a reliable and monotone readiness signal rests on a single illustrative example. Provide aggregate evidence over the test set: for example, the distribution of the divergence at the true aligned read position versus one step earlier, or the correlation between the divergence and the latency penalty of writing at that position. Without such evidence, the choice of δ remains arbitrary and the failure mode seen in the w/o Range ablation is unsurprising.
minor comments (5)
  1. [System Settings / Experiment Details] The paper calls the method 'off-the-shelf' but for text tasks fine-tunes Llama2-7B-chat with LoRA on 100k samples; please clarify that the framework is applicable to off-the-shelf models but the reported text experiments use fine-tuned adapters.
  2. [Table 3] The segment-size ablation shows two rows for 320 ms and 640 ms but only one row for 960 ms; report the low-latency operating point for 960 ms as well, or explain why it is absent.
  3. [Figure 3(c)] The legend uses 'Diseg' while the text uses 'DiSeg'; please ensure consistent naming.
  4. [Evaluation] The BLEU differences in Figure 3 are not accompanied by significance tests or confidence intervals; since greedy decoding is deterministic, a bootstrap over test sentences would strengthen the claim that the improvements are not due to test-set noise.
  5. [Quality of Policy] The sufficiency rate depends on eflomal alignments; alignment errors could bias the comparison, so please consider reporting agreement with a second aligner or providing alignment-quality statistics.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the LSG policy is an empirical heuristic with hand-set operating points, not a result that reduces to its own inputs or to a load-bearing self-citation.

full rationale

The paper's central claim is that an off-the-shelf LLM can act as both policy-maker and generator via a KL-divergence readiness test (Eq. 3), a confidence condition (Eq. 4), and a range constraint (Eq. 5). None of these steps is circular. Eq. 3 compares two next-token distributions; it does not define the resulting policy to be the wait-1 baseline, and the policy it induces is not, by construction, equal to the output metric. Eq. 3 is an empirical heuristic supported by a single illustrative example (Figure 1), but weak support is not circularity. The range constraint Eq. 5 bounds the LLM's write position, and Table 2's 'w/o Range' row shows the unconstrained rule performs poorly; however, this is an ablation demonstrating that the hand-set window is important, not a self-definitional reduction. The paper does not present Eqs. 3-5 as a theorem derived from first principles, nor does it claim the (δ, α, L, U) operating points are predicted rather than selected. The latency-quality curves in Tables 5-7 and Figure 3 are reported at chosen hyperparameter settings, which is standard evaluation; tuning free parameters is not the same as fitting a parameter and renaming it a prediction. The external baselines (HMT, Agent-SiMT, StreamSpeech, Wav2Vec2, Whisper) are evaluated independently; no benchmark score is computed from the method's own assumptions. Self-citations to prior work (Agent-SiMT, HMT, StreamSpeech, BayLing) are used as baselines or model choices, not as a load-bearing uniqueness theorem and not as the source of the KL heuristic. The sufficiency-rate analysis uses alignments from an external tool and is a post-hoc metric rather than an objective that defines the policy. Therefore no circular step can be quoted and exhibited, and the appropriate finding is no significant circularity.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The method relies on a small set of hand-set thresholds and domain assumptions. The free parameters are the read/write thresholds and search-range bounds, all selected per dataset. The core assumption that distribution difference tracks information sufficiency is empirically motivated but not formally derived. No invented entities are introduced.

free parameters (6)
  • δ (KL threshold) = 9.0 (De-En), 7.5 (En-De), 7.0 (Fr-En)
    Controls the READ/WRITE switch by comparing current vs. wait-1 baseline distribution; tuned per dataset with no sensitivity analysis.
  • α (confidence threshold) = 0.6 (De-En), 0.6 (En-De), 0.5 (Fr-En)
    Triggers WRITE when the LLM's max token probability exceeds α, even if KL is below δ; tuned per dataset.
  • L (pre-read length) = 1, 3, 5, 7 (text); 1, 2, 3, 5 (speech)
    Sets the lower bound of the allowed policy range [L+i-1, L+i-1+U]; different values produce the latency-quality frontier.
  • U (autonomy range) = 4 or 6
    Sets how far beyond the wait-like baseline the LLM may delay a WRITE; chosen with L to cover latency scenarios.
  • Speech segment size = 640 ms
    Selected after ablating 320/640/960 ms; balances latency, quality, and inference cost.
  • LoRA rank and fine-tuning budget = r=8, lr=1e-4, batch 128, 100k samples
    Training hyperparameters for the text LLM; not part of the policy but needed to reproduce the claimed results.
assumptions (4)
  • domain assumption KL divergence between current and wait-1 baseline next-token distributions reliably indicates when enough source content has been read.
    Introduced in preliminary analysis (Figure 1) and formalized in Eq. (3); no formal derivation.
  • domain assumption LLM confidence reflects generation credibility and can be used to override small distribution differences.
    Used in Eq. (4), supported only by a citation to Li et al. (2023).
  • domain assumption The optimal generation time for each target word lies within the fixed range [L+i-1, min(L+i-1+U, J)].
    Stated as Eq. (5); the paper motivates it by avoiding outlier policies but gives no evidence that the range always contains the optimal policy.
  • standard math Sufficiency rate computed from eflomal word alignments is a valid measure of policy quality.
    The appendix defines Eq. (6); relies on alignment quality of the external tool.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Large Language Models Are Read/Write Policy-Makers for Simultaneous Generation." pith.science (2026). https://pith.science/paper/LIZ2WLZV

@misc{pith2026250100868,
  author       = {Pith},
  title        = {Pith review of: Large Language Models Are Read/Write Policy-Makers for Simultaneous Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LIZ2WLZV}},
  note         = {Machine review of arXiv:2501.00868}
}
read the original abstract

Simultaneous generation models write generation results while reading streaming inputs, necessitating a policy-maker to determine the appropriate output timing. Existing simultaneous generation methods generally adopt the traditional encoder-decoder architecture and learn the generation and policy-making capabilities through complex dynamic programming techniques. Although LLMs excel at text generation, they face challenges in taking on the role of policy-makers through traditional training methods, limiting their exploration in simultaneous generation. To overcome these limitations, we propose a novel LLM-driven Simultaneous Generation (LSG) framework, which allows the off-the-shelf LLM to decide the generation timing and produce output concurrently. Specifically, LSG selects the generation policy that minimizes latency as the baseline policy. Referring to the baseline policy, LSG enables the LLM to devise an improved generation policy that better balances latency and generation quality, and writes generation results accordingly. Experiments on simultaneous translation and streaming automatic speech recognition tasks show that our method can achieve state-of-the-art performance utilizing the open-source LLMs and demonstrate practicality in real-world scenarios.

Figures

Figures reproduced from arXiv: 2501.00868 by the authors.

Figure 1
Figure 1. The distribution difference of subsequent genera [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework of LLM-driven Simultaneous Generation Model. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Performance of simultaneous generation models on De [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: The performance of LSG framework when em [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Comparison of the policy sufficiency of different [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: The results of LSG method on common voice fr 19617179 from the CoVoST2 Fr-En test set. Transcription: Il séjourne pendant trois ou quatre ans à Paris et Versailles. Reference: He stayed for three or four years at Paris and Versailles. Translation: He stayed there for t…
Figure 7
Figure 7. Figure 7: The results of LSG method on common voice fr 19439076 from the CoVoST2 Fr-En test set. Template: Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.\n\n###Instruct…
Figure 8
Figure 8. Figure 8: The prompt template of LLMs on the SimulT2TT task, where ‘input’ denotes the partial source sentence and ‘output’ [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: The prompt template of LLMs on the SimulS2TT task, where ‘audio’ denotes the input audio. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: The prompt template of LLMs on the Streaming ASR task, where ‘audio’ denotes the input audio. [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SimulS2ST-Omni: Data-Efficient Streaming Speech-to-Speech Translation via Explicit Trajectory Supervision

    cs.SD 2026-07 conditional novelty 6.0 of 10

    A joint text-code trajectory supervision recipe lets a two-stream speech LM achieve competitive long-form streaming S2ST with ~2k hours of paired speech.

  2. StreamUni: Achieving Streaming Speech Translation with a Unified Large Speech-Language Model

    cs.CL 2025-07 conditional novelty 6.0 of 10

    A unified large speech-language model uses speech chain-of-thought to jointly perform segmentation, generation-policy decisions, and streaming translation.

  3. Overcoming Non-monotonicity in Transducer-based Streaming Generation

    cs.CL 2024-11 conditional novelty 6.0 of 10

    MonoAttn-Transducer trains Transducers with monotonic cross-attention estimated from forward-backward posterior alignments, improving simultaneous speech translation quality at similar latency.

Reference graph

Works this paper leans on

54 extracted references · 35 canonical work pages · cited by 3 Pith papers

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Agostinelli, V.; Wild, M.; Raffel, M.; Fuad, K. A. A.; and Chen, L. 2024. Simul-LLM: A Framework for Exploring High-Quality Simultaneous Translation with Large Language Models. arXiv:2312.04691

    arXiv 2024

  4. [4]

    S.; and Sarkar, A

    Alinejad, A.; Shavarani, H. S.; and Sarkar, A. 2021. Translation-based Supervision for Policy Generation in Simultaneous Neural Machine Translation. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, 1734--1744. Online and Punta Cana, Dominican Republic: ...

    work page 2021

  5. [5]

    Arivazhagan, N.; Cherry, C.; Macherey, W.; Chiu, C.-C.; Yavuz, S.; Pang, R.; Li, W.; and Raffel, C. 2019. Monotonic Infinite Lookback Attention for Simultaneous Machine Translation. In Korhonen, A.; Traum, D.; and M \`a rquez, L., eds., Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, 1313--1323. Florence, Italy: As...

    work page 2019

  6. [6]

    Baevski, A.; Zhou, H.; Mohamed, A.; and Auli, M. 2020. wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations. arXiv:2006.11477

    arXiv 2020

  7. [7]

    Bai, J.; Bai, S.; Chu, Y.; Cui, Z.; Dang, K.; Deng, X.; Fan, Y.; Ge, W.; Han, Y.; Huang, F.; Hui, B.; Ji, L.; Li, M.; Lin, J.; Lin, R.; Liu, D.; Liu, G.; Lu, C.; Lu, K.; Ma, J.; Men, R.; Ren, X.; Ren, X.; Tan, C.; Tan, S.; Tu, J.; Wang, P.; Wang, S.; Wang, W.; Wu, S.; Xu, B.; Xu, J.; Yang, A.; Yang, H.; Yang, J.; Yang, S.; Yao, Y.; Yu, B.; Yuan, H.; Yuan,...

    arXiv 2023

  8. [8]

    Chen, J.; Ma, M.; Zheng, R.; and Huang, L. 2021 a . Direct Simultaneous Speech-to-Text Translation Assisted by Synchronized Streaming ASR . In Zong, C.; Xia, F.; Li, W.; and Navigli, R., eds., Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, 4618--4624. Online: Association for Computational Linguistics

    work page 2021

Show all 54 references
  1. [9]

    Chen, J.; Zheng, R.; Kita, A.; Ma, M.; and Huang, L. 2021 b . Improving Simultaneous Translation by Incorporating Pseudo-References with Fewer Reorderings. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical Methods ...

    2021

  2. [10]

    Chu, Y.; Xu, J.; Zhou, X.; Yang, Q.; Zhang, S.; Yan, Z.; Zhou, C.; and Zhou, J. 2023. Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models. arXiv:2311.07919

    2023 arXiv

  3. [11]

    A.; Cattoni, R.; Bentivogli, L.; Negri, M.; and Turchi, M

    Di Gangi, M. A.; Cattoni, R.; Bentivogli, L.; Negri, M.; and Turchi, M. 2019. M u ST - C : a M ultilingual S peech T ranslation C orpus. In Burstein, J.; Doran, C.; and Solorio, T., eds., Proceedings of the 2019 Conference of the North A merican Chapter of the Association for ...

    2019

  4. [12]

    Dong, Q.; Zhu, Y.; Wang, M.; and Li, L. 2022. Learning When to Translate for Streaming Speech. In Muresan, S.; Nakov, P.; and Villavicencio, A., eds., Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 680--694. Dub...

    2022

  5. [13]

    Elbayad, M.; Besacier, L.; and Verbeek, J. 2020. Efficient wait-k models for simultaneous machine translation. arXiv preprint arXiv:2005.08595

    2020 arXiv

  6. [14]

    Fan, R.; Zhou, P.; Chen, W.; Jia, J.; and Liu, G. 2019. An Online Attention-Based Model for Speech Recognition . In Proc. Interspeech 2019, 4390--4394

    2019

  7. [15]

    Gu, J.; Neubig, G.; Cho, K.; and Li, V. O. 2017. Learning to Translate in Real-time with Neural Machine Translation. In Proceedings of the 15th Conference of the E uropean Chapter of the Association for Computational Linguistics: Volume 1, Long Papers , 1053--1062. Valencia, S...

    2017

  8. [16]

    Guo, S.; Zhang, S.; and Feng, Y. 2023 a . Learning Optimal Policy for Simultaneous Machine Translation via Binary Search. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics

    2023

  9. [17]

    Guo, S.; Zhang, S.; and Feng, Y. 2023 b . Simultaneous Machine Translation with Tailored Reference. In Bouamor, H.; Pino, J.; and Bali, K., eds., Findings of the Association for Computational Linguistics: EMNLP 2023, 3070--3084. Singapore: Association for Computational Linguistics

    2023

  10. [18]

    Guo, S.; Zhang, S.; and Feng, Y. 2024 a . Decoder-only Streaming Transformer for Simultaneous Translation. arXiv:2406.03878

    2024 arXiv

  11. [19]

    Guo, S.; Zhang, S.; and Feng, Y. 2024 b . Glancing Future for Simultaneous Machine Translation. In ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 11386--11390

    2024

  12. [20]

    Guo, S.; Zhang, S.; Ma, Z.; Zhang, M.; and Feng, Y. 2024. Agent-SiMT: Agent-assisted Simultaneous Machine Translation with Large Language Models. arXiv:2406.06910

    2024 arXiv

  13. [21]

    J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W

    Hu, E. J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W. 2021. LoRA: Low-Rank Adaptation of Large Language Models. arXiv:2106.09685

    2021 arXiv

  14. [22]

    Jiao, W.; Huang, J.-t.; Wang, W.; He, Z.; Liang, T.; Wang, X.; Shi, S.; and Tu, Z. 2023. P arro T : Translating during Chat using Large Language Models tuned with Human Translation and Feedback. In Bouamor, H.; Pino, J.; and Bali, K., eds., Findings of the Association for Comp...

    2023

  15. [23]

    Koshkin, R.; Sudoh, K.; and Nakamura, S. 2024. TransLLaMa: LLM-based Simultaneous Translation System. arXiv:2402.04636

    2024 arXiv

  16. [24]

    N.; Pang, R.; He, Y.; Strohman, T.; and Wu, Y

    Li, B.; Chang, S.-y.; Sainath, T. N.; Pang, R.; He, Y.; Strohman, T.; and Wu, Y. 2020. Towards Fast and Accurate Streaming End-To-End ASR. In ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 6069--6073

    2020

  17. [25]

    L.; Holtzman, A.; Fried, D.; Liang, P.; Eisner, J.; Hashimoto, T.; Zettlemoyer, L.; and Lewis, M

    Li, X. L.; Holtzman, A.; Fried, D.; Liang, P.; Eisner, J.; Hashimoto, T.; Zettlemoyer, L.; and Lewis, M. 2023. Contrastive Decoding: Open-ended Text Generation as Optimization. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N., eds., Proceedings of the 61st Annual Meeting of the...

    2023

  18. [26]

    Liu, D.; Du, M.; Li, X.; Li, Y.; and Chen, E. 2021. Cross Attention Augmented Transducer Networks for Simultaneous Translation. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processi...

    2021

  19. [27]

    Ma, M.; Huang, L.; Xiong, H.; Zheng, R.; Liu, K.; Zheng, B.; Zhang, C.; He, Z.; Liu, H.; Li, X.; Wu, H.; and Wang, H. 2019. STACL: Simultaneous Translation with Implicit Anticipation and Controllable Latency using Prefix-to-Prefix Framework. In Korhonen, A.; Traum, D. R.; and ...

    2019

  20. [28]

    J.; Wang, C.; Gu, J.; and Pino, J

    Ma, X.; Dousti, M. J.; Wang, C.; Gu, J.; and Pino, J. 2020 a . SIMULEVAL : An Evaluation Toolkit for Simultaneous Translation. In Liu, Q.; and Schlangen, D., eds., Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, 14...

    2020

  21. [29]

    Ma, X.; Pino, J.; and Koehn, P. 2020. SimulMT to SimulST: Adapting Simultaneous Text Translation to End-to-End Simultaneous Speech Translation. arXiv:2011.02048

    2020 arXiv

  22. [30]

    M.; Cross, J.; Puzon, L.; and Gu, J

    Ma, X.; Pino, J. M.; Cross, J.; Puzon, L.; and Gu, J. 2020 b . Monotonic Multihead Attention. In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020 . OpenReview.net

    2020

  23. [31]

    Ma, Z.; Fang, Q.; Zhang, S.; Guo, S.; Feng, Y.; and Zhang, M. 2024. A Non-autoregressive Generation Framework for End-to-End Simultaneous Speech-to-Any Translation. In Ku, L.-W.; Martins, A.; and Srikumar, V., eds., Proceedings of the 62nd Annual Meeting of the Association for...

    2024

  24. [32]

    Ma, Z.; Feng, Y.; and Zhang, M. 2024. Learning Monotonic Attention in Transducer for Streaming Generation. arXiv:2411.17170

    2024 arXiv

  25. [33]

    Ma, Z.; Zhang, S.; Guo, S.; Shao, C.; Zhang, M.; and Feng, Y. 2023. Non-autoregressive Streaming Transformer for Simultaneous Translation. In Bouamor, H.; Pino, J.; and Bali, K., eds., Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, 5177...

    2023

  26. [34]

    Miao, Y.; Blunsom, P.; and Specia, L. 2021. A Generative Framework for Simultaneous Machine Translation. In Moens, M.; Huang, X.; Specia, L.; and Yih, S. W., eds., Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021, Virtual Event...

    2021

  27. [35]

    Moritz, N.; Hori, T.; and Le, J. 2020. Streaming Automatic Speech Recognition with the Transformer Model. In ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 6074--6078

    2020

  28. [36]

    Moritz, N.; Hori, T.; and Roux, J. L. 2019. Triggered Attention for End-to-end Speech Recognition. In ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 5666--5670

    2019

  29. [37]

    Papi, S.; Negri, M.; and Turchi, M. 2023. Attention as a Guide for Simultaneous Speech Translation. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N., eds., Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 13340--13...

    2023

  30. [38]

    Post, M. 2018. A Call for Clarity in Reporting BLEU Scores. In Bojar, O.; Chatterjee, R.; Federmann, C.; Fishel, M.; Graham, Y.; Haddow, B.; Huck, M.; Yepes, A. J.; Koehn, P.; Monz, C.; Negri, M.; N \'e v \'e ol, A.; Neves, M.; Post, M.; Specia, L.; Turchi, M.; and Verspoor, K...

    2018

  31. [39]

    W.; Xu, T.; Brockman, G.; McLeavey, C.; and Sutskever, I

    Radford, A.; Kim, J. W.; Xu, T.; Brockman, G.; McLeavey, C.; and Sutskever, I. 2022. Robust Speech Recognition via Large-Scale Weak Supervision. arXiv:2212.04356

    2022 arXiv

  32. [40]

    C.; and Lavie, A

    Rei, R.; Stewart, C.; Farinha, A. C.; and Lavie, A. 2020. COMET : A Neural Framework for MT Evaluation. In Webber, B.; Cohn, T.; He, Y.; and Liu, Y., eds., Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2685--2702. Online: Assoc...

    2020

  33. [41]

    Tang, Y.; Sun, A.; Inaguma, H.; Chen, X.; Dong, N.; Ma, X.; Tomasello, P.; and Pino, J. 2023. Hybrid Transducer and Attention based Encoder-Decoder Modeling for Speech-to-Text Tasks. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N., eds., Proceedings of the 61st Annual Meeting ...

    2023

  34. [42]

    Touvron, H.; Martin, L.; Stone, K.; Albert, P.; Almahairi, A.; Babaei, Y.; Bashlykov, N.; Batra, S.; Bhargava, P.; Bhosale, S.; Bikel, D.; Blecher, L.; Ferrer, C. C.; Chen, M.; Cucurull, G.; Esiobu, D.; Fernandes, J.; Fu, J.; Fu, W.; Fuller, B.; Gao, C.; Goswami, V.; Goyal, N....

    2023 arXiv

  35. [43]

    N.; Kaiser, L.; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, L.; and Polosukhin, I. 2017. Attention is All you Need. In Guyon, I.; von Luxburg, U.; Bengio, S.; Wallach, H. M.; Fergus, R.; Vishwanathan, S. V. N.; and Garnett, R., eds., Advances in Neura...

    2017

  36. [44]

    Wang, C.; Wu, A.; and Pino, J. 2020. CoVoST 2 and Massively Multilingual Speech-to-Text Translation. arXiv:2007.10310

    2020 arXiv

  37. [45]

    Yeh, C.-F.; Mahadeokar, J.; Kalgaonkar, K.; Wang, Y.; Le, D.; Jain, M.; Schubert, K.; Fuegen, C.; and Seltzer, M. L. 2019. Transformer-Transducer: End-to-End Speech Recognition with Self-Attention. arXiv:1910.12977

    2019 arXiv

  38. [46]

    Yu, D.; Kang, X.; Liu, Y.; Zhou, Y.; and Zong, C. 2024. Self-Modifying State Modeling for Simultaneous Machine Translation. arXiv:2406.02237

    2024 arXiv

  39. [47]

    Zeng, X.; Li, L.; and Liu, Q. 2021. R eal T ran S : End-to-End Simultaneous Speech Translation with Convolutional Weighted-Shrinking Transformer. In Zong, C.; Xia, F.; Li, W.; and Navigli, R., eds., Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, 24...

    2021

  40. [48]

    Zhang, S.; Fang, Q.; Guo, S.; Ma, Z.; Zhang, M.; and Feng, Y. 2024. StreamSpeech: Simultaneous Speech-to-Speech Translation with Multi-task Learning. arXiv:2406.03049

    2024 arXiv

  41. [49]

    Zhang, S.; Fang, Q.; Zhang, Z.; Ma, Z.; Zhou, Y.; Huang, L.; Bu, M.; Gui, S.; Chen, Y.; Chen, X.; and Feng, Y. 2023. BayLing: Bridging Cross-lingual Alignment and Instruction Following through Interactive Translation for Large Language Models. arXiv:2306.10968

    2023 arXiv

  42. [50]

    Zhang, S.; and Feng, Y. 2022 a . Information-Transport-based Policy for Simultaneous Translation. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, 992--1013. Abu Dhabi, United Arab Emirates: Association for Computational Linguistics

    2022

  43. [51]

    Zhang, S.; and Feng, Y. 2022 b . Modeling Dual Read/Write Paths for Simultaneous Machine Translation. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2461--2477. Dublin, Ireland: Association for Computational ...

    2022

  44. [52]

    Zhang, S.; and Feng, Y. 2023 a . End-to-End Simultaneous Speech Translation with Differentiable Segmentation. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N., eds., Findings of the Association for Computational Linguistics: ACL 2023, 7659--7680. Toronto, Canada: Association fo...

    2023

  45. [53]

    Zhang, S.; and Feng, Y. 2023 b . Hidden Markov Transformer for Simultaneous Machine Translation. CoRR, abs/2303.00257

    2023 arXiv

  46. [54]

    Zhang, S.; Guo, S.; and Feng, Y. 2022. Wait-info Policy: Balancing Source and Target at Information Level for Simultaneous Machine Translation. In Goldberg, Y.; Kozareva, Z.; and Zhang, Y., eds., Findings of the Association for Computational Linguistics: EMNLP 2022, 2249--2263...

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.