Predicting Compact Phrasal Rewrites with Large Language Models for ASR Post Editing

Felix Stahlberg; Hao Zhang; Shankar Kumar

REVIEW 4 major objections 5 minor 18 references

Predicting Compact Phrasal Rewrites with Large Language Models for ASR Post Editing

T0 review · 4 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 A target-phrase-only edit representation gives LLM-based ASR post-editing its best efficiency-accuracy trade-off, closing 50-60% of the WER gap between span edits and full rewrites while losing only 10-20% of the length reduction.

desk verdict A clean, modest paper: the target-phrase-only edit representation is a genuine improvement over numeric span indices for LLM-based ASR post editing, and the reported numbers are consistent. read the letter →

arxiv 2501.13831 v1 pith:KUXANMWE submitted 2025-01-23 cs.CL cs.AIcs.LG

Hao Zhang , Felix Stahlberg , Shankar Kumar This is my paper · ORCID

classification cs.CLcs.AIcs.LG

keywords ASRposteditingcompacteditrepresentationstarget-phrase-onlyrepresentationspanLLMrewritingdecodingefficiencyoutputlengthreductionworderrorrate

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 tries to show that an LLM can correct ASR transcripts accurately and cheaply if, instead of rewriting the whole transcript, it only writes the replacement phrases, each surrounded by a few anchor words copied from the input. The compact representation is built from a Levenshtein alignment between the ASR output and the reference, and the LLM is fine-tuned to predict the compressed edit string. On the LibriSpeech test sets, the target-phrase-only representation reduces the average decoded output length by about 56-70 percent while closing 54-57 percent of the word-error-rate gap between the very short edit-span representation and the full-rewrite model. This matters because LLM decoding cost grows with the number of generated tokens, so a shorter output directly means cheaper and faster post-editing.

What carries the argument

The load-bearing object is a compressed edit string built by the compression function $C$ and expanded by the expansion function $E$, which together satisfy $E(x, C(x,y)) = y$. For each edit, the string contains a target phrase extended left and right by $k$ anchor tokens copied from the input, prefixed by a prompt word like 'rewrite'. The anchor tokens are matched against the input during expansion, and the target phrase replaces the matched span; with $k=3$ the anchors are long enough to identify the correct span in almost all dev examples. The dilation parameter $k$ is the control knob that trades output length against ambiguity, and $k=3$ is the setting the paper identifies as the best trade-off across two model sizes.

What would settle it

A direct check is to re-expand the same decoded target-only outputs on the LibriSpeech test sets using a rightmost-match tie-break instead of the leftmost-and-closest rule and compare WER; if the 50-60% gap closure mostly disappears, the reported trade-off is an artifact of the tie-break choice rather than of the representation.

Watch

Extended reading notes

Core claim

The central claim is that the target-phrase-only edit string, with dilation span $k=3$, offers the best point on the efficiency-accuracy curve for LLM-based ASR post-editing. Each edit is written as a target phrase with three words of context on each side; those context words act as anchors that are matched against the input, and the matched span is replaced. The paper reports that this representation recovers the intended rewrite for 99.8% of dev examples, and on the LibriSpeech test sets it lands between the span and full-rewrite extremes: WER of 3.0 on test-clean and 6.8 on test-other, compared with 2.7 and 6.2 for full rewrites and 3.4 and 7.5 for span edits, with average output lengths of 6 and 8 tokens versus 20 and 18 for full rewrites. That is what the 50-60% gap closure and 10-20% length-loss numbers describe.

Load-bearing premise

The load-bearing premise is that the anchor words copied around a target phrase will identify the one place in the transcript where that phrase should be applied; the paper's expansion stage enforces this with a leftmost-and-closest tie-break when several places match, and the reported gains depend on that rule picking correctly.

Editorial extensions

If this is right

LLM-based ASR post-editing can run with about a third to under half of the decoded output length of a full rewrite while keeping most of the WER gain.
The same representation works for both smaller and larger decoder-only LLMs, since the trade-off curve is consistent for the two model sizes tested.
Because the expansion step is a deterministic string-matching and replacement algorithm, the post-editing system does not need a second model to decode or reconstruct the final transcript.
The phrase-pair representation recovers the intended rewrite in 99.98% of dev examples, the highest recovery rate among the phrasal representations, making it the safer choice when reconstruction fidelity is the priority.

Reading between the lines

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

The paper does not ablate the leftmost-and-closest tie-break rule used when anchor words match the input in more than one place, so the reported 50-60% gap closure is conditional on that rule; varying the tie-break is the first testable check.
The same compression-expansion framework could be applied to other high-overlap rewriting tasks, such as grammatical error correction or style transfer, and the recovery-rate table suggests the dilation size would need to be tuned per task.
A natural next experiment is asymmetric dilation, giving more anchor context on the left than on the right, since left context may be more informative for a left-to-right decoder; the paper notes this possibility but does not test it.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Worth a read if you care about compact edit representations for LLM rewriting. The target-only representation — replacing numeric span indices with natural-language anchor phrases — is a simple and sensible idea, and the authors evaluate it carefully against the span baseline and a full-rewrite baseline on LibriSpeech. The arithmetic holds up: the abstract's 50-60% WER-gap closure and 10-20% length-reduction loss are direct from Table II. The paper is clearly written, the expansion functions are defined precisely, and the Limitations section is candid about edit-distance alignment and uniform dilation.

The soft spots are real but mostly minor. The efficiency claim is about output length, not measured latency. That's standard in this line of work, but it means the '70% cost cut' is an estimate, not a benchmark. There are no error bars or significance tests; with 5-point WER differences and no variance, I'd want a sanity check on whether the target-only vs phrase-pair ordering is stable. The bigger question is the unablated tie-break in the target-only expansion. The stress-test concern is fair in principle — leftmost-and-closest is a heuristic — but the paper reports a 99.8% recovery rate on dev for k=3, which directly measures how often the gold compression survives expansion. That makes the tie-break a minor worry, not a load-bearing flaw. It would still be worth a one-line ablation or a test-set recovery number.

The lack of code and data is a real limitation at this venue; the models are proprietary, so nobody can reproduce the WER numbers independently. That is a moderate blow to confidence, not a reason to reject.

Overall, this is a modest, honest contribution. The target-only representation is new, the comparison is systematic, and the central claim is consistent with the reported numbers. I'd send it to reviewers — they'll want the latency measurement and a tie-break ablation, but the core idea deserves a proper hearing.

Referee Report

4 major / 5 minor

Summary. The paper proposes two compact edit-representation schemes for LLM-based rewriting applied to ASR post editing: a phrase-pair representation (source phrase plus target phrase with dilation context) and a target-phrase-only representation (dilated target phrase with anchor context, located by deterministic string matching). Both are compared against the edit-span representation of Kaneko and Okazaki (2023) and against a full-rewrite baseline. The authors fine-tune PaLM 2 Gecko and Otter models on LibriSpeech, freeze the USM ASR model, and report WER and decoder output length on test-clean and test-other. They claim that the target-phrase-only representation offers the best efficiency-accuracy trade-off, closing 50-60% of the WER gap between span and full rewrite while losing only 10-20% of the span model's length reduction rate. The paper also reports dev-set recovery rates for the expansion step and discusses limitations, including the use of edit-distance-based phrases and uniform dilation spans.

Significance. If the trade-off claim holds, the target-only representation is a practical way to reduce autoregressive decoding cost for ASR post editing with modest WER degradation, and the paper would be a useful extension of the edit-span idea to phrasal, more natural edit formats. The arithmetic in Table II is internally consistent: the 57% and 54% WER-gap closures and the 12.5% and 22.2% length-reduction losses follow directly from the reported numbers. The paper also merits credit for selecting k on dev and reporting test WER and length, for comparing two model sizes, and for including an explicit limitations section. However, the central efficiency claim is supported only by output-length reduction, not by measured latency or throughput, and the deterministic expansion heuristic is not ablated, so the robustness of the reported gap closure is not fully established.

major comments (4)

[Section III-A; Table II] The efficiency axis is decoder output length reduction, but the abstract and conclusion frame the contribution in terms of computational cost ('decoding cost still increases with output length') and even 'slowing down decoding by 10-20%'. No wall-clock latency, throughput, or FLOP measurements are reported. Since the Limitations section states that the target-only expansion is 'more algorithmically involved' than the other two compact representations, the real end-to-end speedup may not be proportional to output length. Please report measured latency/throughput for the full pipeline, or explicitly restrict the efficiency claims to output-length reduction.
[Abstract; Section III; Table II] The abstract claims the method 'loses only 10-20% of the length reduction rate of the edit span model.' On test-other, the reported numbers are span = 5 tokens (-72%) and target-only = 8 tokens (-56%), a relative loss of (72-56)/72 = 22.2%, which the main text itself states. The abstract should be corrected or qualified (e.g., '10-20% on test-clean').
[Section III-B; Table III; Section II-C] The reported recovery rate is computed on the reference compression C(x,y) from the dev set, not on LLM-predicted edits, and no test-set recovery rate is given. The WER numbers on test are end-to-end outcomes and therefore already include any expansion failures, but they do not separate LLM prediction error from expansion-rule error. Please report recovery rates on predicted edits for test-clean and test-other, and ablate the leftmost-and-closest tie-break rule (e.g., rightmost match, longest-anchor match, or drop-on-ambiguity) to show that the 50-60% gap closure is a property of the target-only representation rather than of one hand-chosen matching heuristic.
[Section II-C; Eq. (4)] The expansion function E for the target-only representation is under-specified. The paper says only that string matching and replacement 'deals with discontiguous dilation spans' and gives one example with a leftmost-and-closest tie-break, but it does not define the matching algorithm, the priority order of the two tie-break criteria, or the behavior when the predicted anchor context appears multiple times or not at all. This makes the method difficult to reproduce and leaves a free parameter unexamined. Please provide a precise algorithm or pseudocode for E.

minor comments (5)

[Table II caption] The percentages in parentheses mix bases: WER reductions are relative to USM, while output length reductions are relative to the full model. Please state the reference value for each percentage in the caption or use a consistent notation.
[Figure 3] The legend labels the series as 'target only (k=1,2,3,4)' but the individual points are not annotated with their k values; since the claim that k=3 is best is load-bearing, please label the points or provide a table with the exact WER/length coordinates.
[Section II-C] The phrase 'leftmost and closest pair' is not precisely defined: in the example, both candidate anchor matches start at the same leftmost position, so the reader cannot tell how the two criteria are ordered. Please define the tie-break rule explicitly.
[Abstract; Section I] The abstract says the paper 'systematically compare[s]' representations, but the experiments cover only ASR post editing (one task) with two model sizes; consider qualifying 'systematically' or adding a sentence in the introduction noting that the comparison is task-specific.
[Section III-B] Table III reports dev-set recovery only; please clarify whether the same recovery behavior was observed on the test sets, or rename the table/heading to indicate that it is dev-only, to avoid implying test-set recovery.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's central efficiency-accuracy comparison is an empirical measurement on test data after dev-based hyperparameter selection, not a quantity forced by construction or by self-citation.

full rationale

The paper's derivation chain is self-contained and empirically grounded. The target-phrase-only representation is defined by the compression function in Eq. 4 and a deterministic expansion function E; no parameter is fitted to the test WER numbers reported in Table II. The dilation size k=3 is selected on the dev set, and the final WER and average output length values for test-clean and test-other are measured after expanding model predictions, so the claimed 50-60% WER-gap closure is a direct arithmetic consequence of Table II (e.g., on test-clean: span WER 3.4, full WER 2.7, target-only WER 3.0, giving (3.4-3.0)/(3.4-2.7)=57% closure) rather than a circular restatement of any input. The recovery rates in Table III are measurements of the reference-level identity E(x,C(x,y))=y on the dev set, not fitted values that produce the test results. The tie-break rule for ambiguous anchor matches is a potential robustness limitation, but an unablated heuristic is not circularity because the reported WER is not defined in terms of that heuristic's success. Citations to previous work, including Kaneko and Okazaki's span representation and the PaLM 2 technical report, are external baselines and model descriptions; they do not smuggle in the paper's target conclusion. No step reduces to its own input by definition, so the appropriate finding is no significant circularity.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central empirical claim rests on a hand-tuned dilation span k, a hand-chosen tie-breaking rule, the output-length-as-efficiency proxy, and the assumption that edit-distance spans capture ASR corrections. The representations themselves do not introduce new entities.

free parameters (2)

Dilation span k = 3 (selected on dev set; tested k=1,2,3,4)
Controls how much left/right context is added to phrase pair and target-only representations (Eqs. 3-4). The reported target-only results use k=3, chosen as the best point on the dev-set WER-versus-length curve in Figure 3, so the main test numbers depend on this hand-tuned choice.
Expansion tie-breaking rule = leftmost and closest match
For ambiguous anchor matches in the target-only representation, Section II-C specifies preferring the leftmost and closest pair. This rule affects recovery rate and WER and was chosen by the authors without ablation.

assumptions (4)

domain assumption Every rewrite (x,y) can be represented as a sequence of local, monotonic edit spans derived from Levenshtein alignment.
Used throughout Section II to define C and E. The paper notes in Section III-B that phrase representations only recover the original y close to 100% of the time (Table III), so this representability is not exact for phrase formats.
domain assumption Autoregressive decoding cost is proportional to output length, making output-length reduction a valid efficiency metric.
Section III defines the efficiency metric as decoder output length reduction rate and does not measure wall-clock time or account for expansion and parsing overhead. This is the basis for the efficiency-accuracy trade-off claim.
domain assumption A fine-tuned PaLM 2 model can reliably generate the structured edit-format strings and integer spans used by the three representations.
All experiments in Section III assume the model emits well-formed edits; format failures would surface as recovery errors. Recovery rate is reported, but format validity is not separated from matching failures.
domain assumption ASR post editing can be modeled as a text rewriting task with high input-output overlap, so edit-distance-based local spans capture the corrections.
The introduction frames ASR post editing as a variant of grammatical error correction, and the Limitations section concedes that Levenshtein-derived phrases are only minimal by edit distance, not linguistically meaningful.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Predicting Compact Phrasal Rewrites with Large Language Models for ASR Post Editing." pith.science (2026). https://pith.science/paper/KUXANMWE

@misc{pith2026250113831,
  author       = {Pith},
  title        = {Pith review of: Predicting Compact Phrasal Rewrites with Large Language Models for ASR Post Editing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KUXANMWE}},
  note         = {Machine review of arXiv:2501.13831}
}

read the original abstract

Large Language Models (LLMs) excel at rewriting tasks such as text style transfer and grammatical error correction. While there is considerable overlap between the inputs and outputs in these tasks, the decoding cost still increases with output length, regardless of the amount of overlap. By leveraging the overlap between the input and the output, Kaneko and Okazaki (2023) proposed model-agnostic edit span representations to compress the rewrites to save computation. They reported an output length reduction rate of nearly 80% with minimal accuracy impact in four rewriting tasks. In this paper, we propose alternative edit phrase representations inspired by phrase-based statistical machine translation. We systematically compare our phrasal representations with their span representations. We apply the LLM rewriting model to the task of Automatic Speech Recognition (ASR) post editing and show that our target-phrase-only edit representation has the best efficiency-accuracy trade-off. On the LibriSpeech test set, our method closes 50-60% of the WER gap between the edit span model and the full rewrite model while losing only 10-20% of the length reduction rate of the edit span model.

Figures

Figures reproduced from arXiv: 2501.13831 by the authors.

**Figure 2.** Overview of the entire ASR system with the edit-representation-based [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** WER versus output length on the dev set. Left: PaLM 2 Gecko model. Right: PaLM 2 Otter model. target only with dilation span size 3 (the third △ from the left) is the best strategy. TABLE III RECOVERY RATE OF PHRASAL REPRESENTATIONS. phrase pair (K=1) AND target only (K=3) HAVE A RECOVERY RATE CLOSE TO 100%. representation recovery rate phrase pair (k = 1) 99.98% target only (k = 1) 96.80% target only (k = 2) 99.50%… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 9 canonical work pages

[1]

Reducing sequence length by predicting edit spans with large language models,

M. Kaneko and N. Okazaki, “Reducing sequence length by predicting edit spans with large language models,” in Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , H. Bouamor, J. Pino, and K. Bali, Eds. Singapore: Association for Computational Linguistics, Dec. 2023, pp. 10 017–10 029. [Online]. Available: https://aclanth...

work page 2023
[2]

Exploring the limits of transfer learning with a unified text-to-text transformer,

C. Raffel, N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y . Zhou, W. Li, and P. J. Liu, “Exploring the limits of transfer learning with a unified text-to-text transformer,” Journal of Machine Learning Research, vol. 21, no. 140, pp. 1–67, 2020. [Online]. Available: http://jmlr.org/papers/v21/20-074.html

work page 2020
[3]

Language models are few-shot learners,

T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert- V oss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amode...

work page 2020
[4]

Palm: Scaling language modeling with pathways,

A. Chowdhery, S. Narang, J. Devlin, M. Bosma, G. Mishra, A. Roberts, P. Barham, H. W. Chung, C. Sutton, S. Gehrmann, P. Schuh, K. Shi, S. Tsvyashchenko, J. Maynez, A. Rao, P. Barnes, Y . Tay, N. Shazeer, V . Prabhakaran, E. Reif, N. Du, B. Hutchinson, R. Pope, J. Bradbury, J. Austin, M. Isard, G. Gur-Ari, P. Yin, T. Duke, A. Levskaya, S. Ghemawat, S. Dev,...
[5]

Palm 2 technical report,

R. Anil, A. M. Dai, O. Firat, M. Johnson, D. Lepikhin, A. Passos, S. Shakeri, E. Taropa, P. Bailey, Z. Chen, E. Chu, J. H. Clark, L. E. Shafey, Y . Huang, K. Meier-Hellstern, G. Mishra, E. Moreira, M. Omernick, K. Robinson, S. Ruder, Y . Tay, K. Xiao, Y . Xu, Y . Zhang, G. H. Abrego, J. Ahn, J. Austin, P. Barham, J. Botha, J. Bradbury, S. Brahma, K. Brook...

arXiv 2023
[6]

Gpt-4 technical report,

OpenAI, J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, R. Avila, I. Babuschkin, S. Balaji, V . Balcom, P. Baltescu, H. Bao, M. Bavarian, J. Belgum, I. Bello, J. Berdine, G. Bernadett-Shapiro, C. Berner, L. Bogdonoff, O. Boiko, M. Boyd, A.-L. Brakman, G. Brockman, T. Brooks, M. Brunda...

arXiv 2024
[7]

A recipe for arbitrary text style transfer with large language models,

E. Reif, D. Ippolito, A. Yuan, A. Coenen, C. Callison-Burch, and J. Wei, “A recipe for arbitrary text style transfer with large language models,” in Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), Dublin, Ireland, May 2022, pp. 837–848. [Online]. Available: https://aclanthology.org/2022.acl...

work page 2022
[8]

A simple recipe for multilingual grammatical error correction,

S. Rothe, J. Mallinson, E. Malmi, S. Krause, and A. Severyn, “A simple recipe for multilingual grammatical error correction,” in Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), Online, Aug. 2021, pp. 702–707. [Online...

work page 2021

Show all 18 references

[9]

Is chatgpt a highly fluent grammatical error correction system? a comprehensive evaluation,

T. Fang, S. Yang, K. Lan, D. F. Wong, J. Hu, L. S. Chao, and Y . Zhang, “Is chatgpt a highly fluent grammatical error correction system? a comprehensive evaluation,” 2023. [Online]. Available: https://arxiv.org/abs/2304.01746

2023 arXiv
[10]

Statistical phrase-based translation,

P. Koehn, F. J. Och, and D. Marcu, “Statistical phrase-based translation,” in Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics , 2003, pp. 127–133. [Online]. Available: https://aclanthol...

2003
[11]

Correcting ESL errors using phrasal SMT techniques,

C. Brockett, W. B. Dolan, and M. Gamon, “Correcting ESL errors using phrasal SMT techniques,” in Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the Association for Computational Linguistics , Sydney, Australia, Jul. 200...

2006
[12]

Google usm: Scaling automatic speech recognition beyond 100 languages,

Y . Zhang, W. Han, J. Qin, Y . Wang, A. Bapna, Z. Chen, N. Chen, B. Li, V . Axelrod, G. Wang, Z. Meng, K. Hu, A. Rosenberg, R. Prabhavalkar, D. S. Park, P. Haghani, J. Riesa, G. Perng, H. Soltau, T. Strohman, B. Ramabhadran, T. Sainath, P. Moreno, C.-C. Chiu, J. Schalkwyk, F. ...

2023 arXiv
[13]

Librispeech: An asr corpus based on public domain audio books,

V . Panayotov, G. Chen, D. Povey, and S. Khudanpur, “Librispeech: An asr corpus based on public domain audio books,” in 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2015, pp. 5206–5210

2015
[14]

Fast inference from transform- ers via speculative decoding,

Y . Leviathan, M. Kalman, and Y . Matias, “Fast inference from transform- ers via speculative decoding,” in Proceedings of the 40th International Conference on Machine Learning , ser. ICML’23. JMLR.org, 2023

2023
[15]

Accelerating large language model decoding with speculative sampling,

C. Chen, S. Borgeaud, G. Irving, J.-B. Lespiau, L. Sifre, and J. Jumper, “Accelerating large language model decoding with speculative sampling,” 2023. [Online]. Available: https://arxiv.org/abs/2302.01318

2023 arXiv
[16]

Multi-stage large language model correction for speech recognition,

J. Pu, T.-S. Nguyen, and S. St ¨uker, “Multi-stage large language model correction for speech recognition,” 2024. [Online]. Available: https://arxiv.org/abs/2310.11532

2024 arXiv
[17]

Fastcorrect: Fast error correction with edit alignment for automatic speech recognition,

Y . Leng, X. Tan, L. Zhu, J. Xu, R. Luo, L. Liu, T. Qin, X. Li, E. Lin, and T.-Y . Liu, “Fastcorrect: Fast error correction with edit alignment for automatic speech recognition,” Advances in Neural Information Processing Systems, vol. 34, pp. 21 708–21 719, 2021

2021
[2023]

Available: https://jmlr.org/papers/v24/22-1144.html

[Online]. Available: https://jmlr.org/papers/v24/22-1144.html

Pith tools

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

[1] [1]

Reducing sequence length by predicting edit spans with large language models,

M. Kaneko and N. Okazaki, “Reducing sequence length by predicting edit spans with large language models,” in Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , H. Bouamor, J. Pino, and K. Bali, Eds. Singapore: Association for Computational Linguistics, Dec. 2023, pp. 10 017–10 029. [Online]. Available: https://aclanth...

work page 2023

[2] [2]

Exploring the limits of transfer learning with a unified text-to-text transformer,

C. Raffel, N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y . Zhou, W. Li, and P. J. Liu, “Exploring the limits of transfer learning with a unified text-to-text transformer,” Journal of Machine Learning Research, vol. 21, no. 140, pp. 1–67, 2020. [Online]. Available: http://jmlr.org/papers/v21/20-074.html

work page 2020

[3] [3]

Language models are few-shot learners,

T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert- V oss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amode...

work page 2020

[4] [4]

Palm: Scaling language modeling with pathways,

A. Chowdhery, S. Narang, J. Devlin, M. Bosma, G. Mishra, A. Roberts, P. Barham, H. W. Chung, C. Sutton, S. Gehrmann, P. Schuh, K. Shi, S. Tsvyashchenko, J. Maynez, A. Rao, P. Barnes, Y . Tay, N. Shazeer, V . Prabhakaran, E. Reif, N. Du, B. Hutchinson, R. Pope, J. Bradbury, J. Austin, M. Isard, G. Gur-Ari, P. Yin, T. Duke, A. Levskaya, S. Ghemawat, S. Dev,...

[5] [5]

Palm 2 technical report,

R. Anil, A. M. Dai, O. Firat, M. Johnson, D. Lepikhin, A. Passos, S. Shakeri, E. Taropa, P. Bailey, Z. Chen, E. Chu, J. H. Clark, L. E. Shafey, Y . Huang, K. Meier-Hellstern, G. Mishra, E. Moreira, M. Omernick, K. Robinson, S. Ruder, Y . Tay, K. Xiao, Y . Xu, Y . Zhang, G. H. Abrego, J. Ahn, J. Austin, P. Barham, J. Botha, J. Bradbury, S. Brahma, K. Brook...

arXiv 2023

[6] [6]

Gpt-4 technical report,

OpenAI, J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, R. Avila, I. Babuschkin, S. Balaji, V . Balcom, P. Baltescu, H. Bao, M. Bavarian, J. Belgum, I. Bello, J. Berdine, G. Bernadett-Shapiro, C. Berner, L. Bogdonoff, O. Boiko, M. Boyd, A.-L. Brakman, G. Brockman, T. Brooks, M. Brunda...

arXiv 2024

[7] [7]

A recipe for arbitrary text style transfer with large language models,

E. Reif, D. Ippolito, A. Yuan, A. Coenen, C. Callison-Burch, and J. Wei, “A recipe for arbitrary text style transfer with large language models,” in Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), Dublin, Ireland, May 2022, pp. 837–848. [Online]. Available: https://aclanthology.org/2022.acl...

work page 2022

[8] [8]

A simple recipe for multilingual grammatical error correction,

S. Rothe, J. Mallinson, E. Malmi, S. Krause, and A. Severyn, “A simple recipe for multilingual grammatical error correction,” in Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), Online, Aug. 2021, pp. 702–707. [Online...

work page 2021