Pith. sign in

REVIEW 4 major objections 5 minor 52 references

Xmodel-1.5: An 1B-scale Multilingual LLM

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Xmodel-1.5, a 1-billion-parameter multilingual language model, claims to outperform the larger PolyLM-1.7B on selected Thai, Arabic, French, and Chinese benchmarks and to achieve state-of-the-art results in Thai.

desk verdict The central multilingual comparison is undermined by duplicated accuracy values in Table 7, though the model release and Thai dataset are real contributions. read the letter →

arxiv 2411.10083 v3 pith:HZFNEDOR submitted 2024-11-15 cs.CL

Wang Qun , Liu Yang , Lin Qingquan , Jiang Ling This is my paper · ORCID
classification cs.CL
keywords multilinguallanguagemodelunigramtokenizerlow-resourcelanguagesThaievaluationdatasetXdata_Thai1Bparametercommonsensereasoninginstructiontuning
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

Xmodel-1.5 is a 1-billion-parameter multilingual language model pretrained on 2 trillion tokens. The paper claims that this compact model, built around a custom 65,280-token unigram tokenizer, outperforms the larger PolyLM-1.7B on selected Thai, Arabic, French, and Chinese evaluation tasks, and that it achieves state-of-the-art results in Thai. It also introduces Xdata_Thai, a 350-question Thai evaluation dataset that highlights challenges like gendered particles and idioms. The authors argue that careful tokenization and targeted multilingual data can make small models competitive in low-resource languages, offering a more scalable path to multilingual AI.

What carries the argument

The load-bearing mechanism is the custom unigram tokenizer. Trained with SentencePiece on a 50GB subset of the pretraining corpus (50% English, 25% Chinese, 10% industry-specific, 15% low-resource languages), it uses byte fallback for rare characters, splits numbers into digits, and keeps extra whitespace instead of removing it. The resulting 65,280-token vocabulary achieves a compression rate of 0.3800, beating LLaMA 3's 0.3823 despite having about half the vocabulary size. The tokenizer's flexibility with low-frequency tokens is what the paper argues lets a 1B model handle Thai and Arabic morphology efficiently enough to outperform a 1.7B model.

What would settle it

Re-run PolyLM-1.7B on the Table 7 tasks under the exact 3-shot, token-based matching protocol used for Xmodel-1.5; if PolyLM-1.7B then matches or exceeds Xmodel-1.5's scores, the outperformance claim is falsified. Alternatively, run every publicly available Thai-capable model on Xdata_Thai under the same 3-shot prompt and show that any model scores above 0.237, falsifying the state-of-the-art-in-Thai claim.

Watch

Extended reading notes

Core claim

The central claim is that Xmodel-1.5-1B beats PolyLM-1.7B on the paper's chosen multilingual benchmarks, including Belebele Thai, XCOPA Thai, Chinese ARC-e, and Arabic PIQA, and achieves state-of-the-art results in Thai. The paper attributes much of this to the custom unigram tokenizer: a 65,280-token SentencePiece unigram vocabulary with byte fallback, digit splitting, and no extra-whitespace removal, which reaches a compression rate of 0.3800, lower than several larger BPE vocabularies. On the released Xdata_Thai benchmark, Xmodel-1.5 scores 0.237 versus PolyLM-1.7B's 0.228 under a 3-shot setting, and the model also posts a 92.47% satisfaction rate on an e-commerce RAG evaluation after instruction tuning.

Load-bearing premise

The load-bearing premise is that the multilingual evaluations used the same prompt format, number of shots, normalization, and token-based matching for Xmodel-1.5 and PolyLM-1.7B; the paper spells out those settings only for its own Xdata_Thai benchmark, not for the Table 7 results.

Editorial extensions

If this is right

  • If correct, the results show that 1-billion-parameter models can compete with 1.7B models on targeted multilingual benchmarks, making deployment cheaper and faster.
  • The unigram tokenizer's compression advantage suggests that tokenization design can matter as much as model scale for low-resource languages.
  • Xdata_Thai provides a reusable benchmark for cultural-linguistic phenomena such as gendered particles and idioms, which standard multilingual benchmarks do not cover.
  • The strong e-commerce RAG performance (92.47% satisfaction) indicates the instruction-tuned model is usable for real commercial multilingual customer service.

Reading between the lines

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

  • The comparison with PolyLM-1.7B could be affected by unspecified evaluation settings for Table 7, so the outperformance should be treated as provisional until the protocol is shared and reproduced.
  • The state-of-the-art in Thai claim rests on comparisons shown in the paper with PolyLM-1.7B and PolyLM-13B; a public leaderboard test against all Thai-capable models would be a stronger check.
  • The same tokenizer recipe could be tested on other low-resource languages, such as Hindi or Swahili, to see whether the compression and accuracy gains generalize.
  • The release of Xdata_Thai, with its focus on idioms and gendered particles, could become a targeted probe for measuring progress on culturally specific language generation.
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

4 major / 5 minor

Summary. The paper introduces Xmodel-1.5, a 1-billion-parameter multilingual language model pretrained on 2 trillion tokens, with a custom 65,280-token unigram tokenizer. The authors describe the data mix, tokenizer design, architecture, and training details, and evaluate the model on English commonsense benchmarks, multilingual tasks (mMMLU, PIQA, XCOPA, Belebele, etc.), and a newly released Thai evaluation dataset, Xdata_Thai. The central claims are that Xmodel-1.5-1B outperforms Alibaba's PolyLM-1.7B on selected Thai, Arabic, French, and Chinese tasks, and that it achieves state-of-the-art results in Thai. The paper also reports instruction-following and chat results, and discusses qualitative feedback from a Chulalongkorn University collaboration.

Significance. If the reported results hold, the paper would provide a useful data point for 1B-scale multilingual models, particularly for low-resource languages like Thai. The public release of model and code, the detailed tokenizer comparison, and the construction of a human-annotated Thai evaluation dataset are concrete contributions. However, the evidence supporting the central empirical claims is currently weakened by apparent data duplications in the main multilingual results table, the absence of a uniform evaluation protocol for the benchmark comparisons, and the lack of statistical grounding for the Thai-specific superiority claim. These issues are fixable but are load-bearing for the paper's conclusions.

major comments (4)
  1. [Section 9.3, Table 7b/7c/7d] Table 7 reports identical accuracy values for three distinct tasks in each model row: PolyLM-1.7B has 0.5381 on piqa_ar, piqa_fr, and xcopa_zh; PolyLM-13B has 0.5653 on those same three tasks; and Xmodel-1.5-1B has 0.5789 on all three. Because these are different datasets (Arabic PIQA, French PIQA, and Chinese XCOPA), four-decimal equality across all three is effectively impossible under independent evaluation. This suggests a copy-paste error or a systematic evaluation bug (e.g., reusing the same prediction file). These rows contribute to the claimed multilingual advantage over PolyLM, so the accuracy values for these tasks must be rerun and corrected, or the affected numbers cannot be considered reliable evidence.
  2. [Section 5.1 and Section 9.3] The evaluation protocol for the multilingual benchmarks in Table 7 and Figure 5 is not specified. The paper gives a detailed protocol only for Xdata_Thai (Appendix 9.2.2: 3-shot continuation prompts, randomized options, first-10-token matching). For mMMLU, PIQA_AR, Belebele, XCOPA, ARC-ZH, and related tasks, the number of shots, prompt formatting (with or without chat template), answer normalization, and whether accuracy is computed by token matching or by log-likelihood are not stated. Without a common protocol across models, the comparison to PolyLM-1.7B is not reliably interpretable, because observed differences could reflect evaluation choices rather than model quality. The authors should specify the exact settings for every task used in the central comparison.
  3. [Abstract, Section 5.1, Section 9.3] The claim of 'state-of-the-art results in Thai' is not supported by the evidence presented. The only Thai comparisons are against PolyLM-1.7B and PolyLM-13B on Xdata_Thai, Belebele_tha, and xcopa_th. No comparison is made to other Thai-capable models (e.g., SeaLLM, Typhoon, WangchanLion, Qwen2.5) or to established Thai benchmarks. The SOTA claim in the abstract and Section 7 should either be backed by a broader comparison or removed.
  4. [Table 6, Appendix 9.2.2] On the 350-sample Xdata_Thai dataset, the reported difference between Xmodel-1.5-1B (0.237) and PolyLM-1.7B (0.228) corresponds to roughly three to four questions, which is within the noise of a 350-item set. The paper reports no confidence intervals, error bars, or significance tests, so the conclusion that Xmodel-1.5 is 'effective' on this dataset and superior to PolyLM is not statistically established. The authors should report the number of correct answers and an appropriate uncertainty quantification or significance test.
minor comments (5)
  1. [Title and Abstract] The phrase 'An 1B-scale' should be 'A 1B-scale' for grammatical correctness.
  2. [Section 5.1, Table 4] The comparison table includes InternLM2-1.8B and Qwen2.5-1.5B, which are 1.8B and 1.5B parameters, respectively, despite the text stating that the baselines have 'approximately 1 billion parameters.' This size mismatch should be acknowledged.
  3. [Section 3.2, Table 2] The compression rate in Table 2 is described as lower-is-better, but the definition of the compression rate (e.g., average tokens per word, bytes per token) is not given. Please add a one-line definition.
  4. [Section 5.1, bullet points] The citation for mHellaswag attributes the benchmark to Hendrycks et al. (2021), which is the MMLU paper; the HellaSwag benchmark is by Zellers et al. (2019). Please correct the reference.
  5. [Appendix 9.2.1] The reference '[Johannes Welbl, 2017]' appears malformed; it should be a proper citation to Welbl et al. (2017).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper reports empirical evaluations against external benchmarks, and the self-citations are provenance notes rather than load-bearing derivations.

full rationale

The paper is an empirical technical report, not a derivation from first principles, so there is no equation-level input/output chain to reduce. The central comparisons (Tables 4, 5, 7; Figures 5-8) use independently published benchmarks (ARC, PIQA, XCOPA, Belebele, mMMLU, mHellaswag, IFEval, MT-Bench) evaluated via the Language Model Evaluation Harness, and the Thai-specific result in Table 6 uses Xdata_Thai, whose 350 answers were annotated by native-speaker students through ad hoc testing, peer review, and consensus selection (Appendix 9.2.1). Self-citations to prior Xmodel work are provenance notes ('Building upon the Chinese and English data from the first generation of Xmodel'; adopting architecture from Xmodel), and generating missing distractors with an earlier XmodelLM is a benchmark-construction detail, not a fitted input or a prediction. The paper's own stated limitations (Section 7; Appendix 9.1 on gendered particles and time/numerical expressions) are empirical weaknesses, not circularity. The abstract's 'state-of-the-art results in Thai' claim lacks a demonstrated comparison to other Thai-specific systems, but that is an evidentiary gap rather than circularity. The identical four-decimal accuracies across piqa_ar, piqa_fr, and xcopa_zh in Table 7 are best explained by a data-handling or reporting error and cast doubt on the PolyLM comparison, but they are not a constructional equivalence. Accordingly, no circular step is identified.

Assumptions & free parameters 0 free parameters · 4 assumptions · 1 invented entities

The central claim is empirical and rests primarily on the fairness and completeness of the evaluation protocol. No mathematical derivation introduces free parameters, but several domain assumptions about the validity of benchmarks and the comparability of baselines are load-bearing.

assumptions (4)
  • domain assumption Transformer-based autoregressive language modeling with the described architecture (RoPE, RMSNorm, SwiGLU, GQA) learns useful multilingual representations from text
    Section 3.3 states architectural choices without justification; the model's performance depends on this standard but unproven assumption.
  • domain assumption The evaluation benchmarks (ARC, HellaSwag, PIQA, mMMLU, XCOPA, Belebele, Xdata_Thai) measure meaningful language capabilities and are not contaminated by the training data
    Section 5 uses these benchmarks as evidence; no contamination analysis is provided.
  • domain assumption The evaluation protocol for multilingual tasks is identical for Xmodel-1.5 and PolyLM baselines
    Section 5.1 and Appendix 9.2.2 do not specify prompt format or normalization for Table 7; the comparison assumes fair conditions.
  • domain assumption The custom unigram tokenizer with 65,280 tokens yields better or equal efficiency than BPE
    Section 3.2 reports compression rates but no downstream ablation tying tokenizer choice to performance.
invented entities (1)
  • Xdata_Thai independent evidence
    purpose: A Thai-specific evaluation dataset with 350 manually annotated samples to benchmark low-resource Thai understanding
    The dataset is released (or planned) publicly; it can be used by others to evaluate Thai models, providing an external falsifiable handle.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Xmodel-1.5: An 1B-scale Multilingual LLM." pith.science (2026). https://pith.science/paper/HZFNEDOR

@misc{pith2026241110083,
  author       = {Pith},
  title        = {Pith review of: Xmodel-1.5: An 1B-scale Multilingual LLM},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HZFNEDOR}},
  note         = {Machine review of arXiv:2411.10083}
}
read the original abstract

We introduce Xmodel-1.5, a 1-billion-parameter multilingual large language model pretrained on 2 trillion tokens, designed for balanced performance and scalability. Unlike most large models that use the BPE tokenizer, Xmodel-1.5 employs a custom unigram tokenizer with 65,280 tokens, optimizing both efficiency and accuracy. The model delivers competitive results across multiple languages, including Thai, Arabic, French, Chinese, and English, outperforming Alibaba's PolyLM-1.7B on respective evaluation datasets. Xmodel-1.5 excels in benchmarks like mMMLU and PIQA, and achieves state-of-the-art results in Thai. To support low-resource language research, we release Xdata_Thai, a Thai-specific evaluation dataset featuring unique linguistic challenges such as gendered particles and idioms. While the model demonstrates strong performance, there is still room for improvement in handling culturally specific nuances. We hope this work contributes to advancements in multilingual AI research. Models and code are publicly available on GitHub at https://github.com/XiaoduoAILab/XmodelLM-1.5

Figures

Figures reproduced from arXiv: 2411.10083 by the authors.

Figure 1
Figure 1. Total multilingual data tokens during the pretraining phase sourced from MultiWiki and [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Data distribution during pretraining be [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. The trend of training and validation loss during pretraining. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figures from the paper (10 more)
Figure 5
Figure 5. Figure 5: Comparison of performance in multilingual tasks between PolyLM 1.7B and Xmodel-1.5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Evolution of performance in Arabic bench during pre-training [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Evolution of performance in Thai bench during pre-training [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Evolution of performance in French bench during pre-training [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Annotation interface for students at Chulalongkorn University [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: The model performs well in e-commerce Q&A, providing concise and clear answers. [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: The model struggles with distinguishing gendered language in Thai, particularly with [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: The model faces challenges in handling time and numerical data in Thai, leading to [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]
Figure 13
Figure 13. Figure 13: The distribution of task types in our evaluation set. Task types with fewer than 5 [PITH_FULL_IMAGE:figures/full_fig_p016_13.png]
Figure 15
Figure 15. Figure 15: Example of a multiple-choice prompt without the chat template. 9.3 Detailed Evaluation Results This section presents a comprehensive evaluation of the model’s performance on multilingual tasks, focusing on Thai, Arabic, French, and Chinese. The evaluation results are …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 16 canonical work pages

  1. [1]

    Gqa: Training generalized multi-query transformer models from multi-head checkpoints, 2023

    Joshua Ainslie, James Lee-Thorp, Michiel de Jong, Yury Zemlyanskiy, Federico Lebrón, and Sumit Sanghai. Gqa: Training generalized multi-query transformer models from multi-head checkpoints, 2023

    2023

  2. [2]

    Wangchanthaiinstruct: A thai instruction-following dataset

    AIRESEARCH. Wangchanthaiinstruct: A thai instruction-following dataset. https://huggingface.co/datasets/airesearch/WangchanThaiInstruct, 2023. URL https://huggingface.co/datasets/airesearch/WangchanThaiInstruct. Accessed: 2024-11-15

    work page 2023

  3. [3]

    A l G hafa evaluation benchmark for A rabic language models

    Ebtesam Almazrouei, Ruxandra Cojocaru, Michele Baldo, Quentin Malartic, Hamza Alobeidli, Daniele Mazzotta, Guilherme Penedo, Giulia Campesan, Mugariya Farooq, Maitha Alhammadi, Julien Launay, and Badreddine Noune. A l G hafa evaluation benchmark for A rabic language models. In Hassan Sawaf, Samhaa El-Beltagy, Wajdi Zaghouani, Walid Magdy, Ahmed Abdelali, ...

    2023

  4. [4]

    Mt-bench-101: A fine-grained benchmark for evaluating large language models in multi-turn dialogues

    Ge Bai, Jie Liu, Xingyuan Bu, Yancheng He, Jiaheng Liu, Zhanhui Zhou, Zhuoran Lin, Wenbo Su, Tiezheng Ge, Bo Zheng, and Wanli Ouyang. Mt-bench-101: A fine-grained benchmark for evaluating large language models in multi-turn dialogues. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), page 7...

    doi:10.18653/v1/2024.acl-long.401 2024

  5. [5]

    The belebele benchmark: a parallel reading comprehension dataset in 122 language variants, 2023

    Lucas Bandarkar, Davis Liang, Benjamin Muller, Mikel Artetxe, Satya Narayan Shukla, Donald Husa, Naman Goyal, Abhinandan Krishnan, Luke Zettlemoyer, and Madian Khabsa. The belebele benchmark: a parallel reading comprehension dataset in 122 language variants, 2023

    work page 2023

  6. [6]

    Pythia: A suite for analyzing large language models across training and scaling, 2023

    Stella Biderman, Hailey Schoelkopf, Quentin Anthony, Herbie Bradley, Kyle O'Brien, Eric Hallahan, Mohammad Aflah Khan, Shivanshu Purohit, USVSN Sai Prashanth, Edward Raff, Aviya Skowron, Lintang Sutawika, and Oskar van der Wal. Pythia: A suite for analyzing large language models across training and scaling, 2023

    2023

  7. [7]

    Piqa: Reasoning about physical commonsense in natural language

    Yonatan Bisk, Rowan Zellers, Ronan Le Bras, Jianfeng Gao, and Yejin Choi. Piqa: Reasoning about physical commonsense in natural language. In AAAI Conference on Artificial Intelligence, 2019. URL https://api.semanticscholar.org/CorpusID:208290939

    2019

  8. [8]

    Byte pair encoding is suboptimal for language model pretraining, 2020

    Kaj Bostrom and Greg Durrett. Byte pair encoding is suboptimal for language model pretraining, 2020. URL https://arxiv.org/abs/2004.03720

    arXiv 2020

Show all 52 references
  1. [9]

    Internlm2 technical report, 2024

    Zheng Cai, Maosong Cao, Haojiong Chen, Kai Chen, Keyu Chen, Xin Chen, Xun Chen, Zehui Chen, Zhi Chen, Pei Chu, Xiaoyi Dong, Haodong Duan, Qi Fan, Zhaoye Fei, Yang Gao, Jiaye Ge, Chenya Gu, Yuzhe Gu, Tao Gui, Aijia Guo, Qipeng Guo, Conghui He, Yingfan Hu, Ting Huang, Tao Jiang,...

    2024

  2. [10]

    Boolq: Exploring the surprising difficulty of natural yes/no questions, 2019

    Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, Michael Collins, and Kristina Toutanova. Boolq: Exploring the surprising difficulty of natural yes/no questions, 2019

    2019

  3. [11]

    Think you have solved question answering? try arc, the ai2 reasoning challenge

    Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. Think you have solved question answering? try arc, the ai2 reasoning challenge. ArXiv, abs/1803.05457, 2018 a

    2018 arXiv

  4. [12]

    Think you have solved question answering? try arc, the ai2 reasoning challenge, 2018 b

    Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. Think you have solved question answering? try arc, the ai2 reasoning challenge, 2018 b

    2018

  5. [13]

    Unsupervised cross-lingual representation learning at scale, 2020

    Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. Unsupervised cross-lingual representation learning at scale, 2020. URL https://arxiv.org/abs/1911.02116

    2020 arXiv

  6. [14]

    wiki-dump-reader

    CyberZHG. wiki-dump-reader. https://github.com/CyberZHG/wiki-dump-reader, 2023. Accessed: 2024-10-23

    2023

  7. [15]

    A framework for few-shot language model evaluation, 12 2023

    Leo Gao, Jonathan Tow, Baber Abbasi, Stella Biderman, Sid Black, Anthony DiPofi, Charles Foster, Laurence Golding, Jeffrey Hsu, Alain Le Noac'h, Haonan Li, Kyle McDonell, Niklas Muennighoff, Chris Ociepa, Jason Phang, Laria Reynolds, Hailey Schoelkopf, Aviya Skowron, Lintang S...

    2023

  8. [16]

    Openwebtext corpus

    Aaron Gokaslan, Vanya Cohen, Ellie Pavlick, and Stefanie Tellex. Openwebtext corpus. http://Skylion007.github.io/OpenWebTextCorpus, 2019

    2019

  9. [17]

    Gordon, Zornitsa Kozareva, and Melissa Roemmele

    Andrew S. Gordon, Zornitsa Kozareva, and Melissa Roemmele. Choice of plausible alternatives: An evaluation of commonsense causal reasoning. In AAAI Spring Symposium: Logical Formalizations of Commonsense Reasoning, 2011. URL https://api.semanticscholar.org/CorpusID:434646

    2011

  10. [18]

    Wanjuan: A comprehensive multimodal dataset for advancing english and chinese large models, 2023

    Conghui He, Zhenjiang Jin, Chao Xu, Jiantao Qiu, Bin Wang, Wei Li, Hang Yan, Jiaqi Wang, and Dahua Lin. Wanjuan: A comprehensive multimodal dataset for advancing english and chinese large models, 2023

    2023

  11. [19]

    Measuring massive multitask language understanding, 2021

    Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. Measuring massive multitask language understanding, 2021. URL https://arxiv.org/abs/2009.03300

    2021 arXiv

  12. [20]

    Qwen2.5-coder technical report, 2024

    Binyuan Hui, Jian Yang, Zeyu Cui, Jiaxi Yang, Dayiheng Liu, Lei Zhang, Tianyu Liu, Jiajun Zhang, Bowen Yu, Kai Dang, An Yang, Rui Men, Fei Huang, Xingzhang Ren, Xuancheng Ren, Jingren Zhou, and Junyang Lin. Qwen2.5-coder technical report, 2024. URL https://arxiv.org/abs/2409.12186

    2024 arXiv

  13. [21]

    Exploring the impact of instruction data scaling on large language models: An empirical study on real-world use cases

    Yunjie Ji, Yong Deng, Yan Gong, Yiping Peng, Qiang Niu, Lei Zhang, Baochang Ma, and Xiangang Li. Exploring the impact of instruction data scaling on large language models: An empirical study on real-world use cases. arXiv preprint arXiv:2303.14742, 2023

    2023 arXiv

  14. [22]

    Matt Gardner Johannes Welbl, Nelson F. Liu. Crowdsourcing multiple choice science questions, 2017

    2017

  15. [23]

    Subword regularization: Improving neural network translation models with multiple subword candidates, 2018 a

    Taku Kudo. Subword regularization: Improving neural network translation models with multiple subword candidates, 2018 a

    2018

  16. [24]

    Subword regularization: Improving neural network translation models with multiple subword candidates, 2018 b

    Taku Kudo. Subword regularization: Improving neural network translation models with multiple subword candidates, 2018 b . URL https://arxiv.org/abs/1804.10959

    2018 arXiv

  17. [25]

    Sentencepiece: A simple and language independent subword tokenizer and detokenizer for neural text processing, 2018

    Taku Kudo and John Richardson. Sentencepiece: A simple and language independent subword tokenizer and detokenizer for neural text processing, 2018

    2018

  18. [26]

    scb-mt-en-th-2020: A large english-thai parallel corpus

    Lalita Lowphansirikul, Charin Polpanumas, Attapol T Rutherford, and Sarana Nutanong. scb-mt-en-th-2020: A large english-thai parallel corpus. arXiv preprint arXiv:2007.03541, 2020

    2020 arXiv

  19. [27]

    Can a suit of armor conduct electricity? a new dataset for open book question answering, 2018

    Todor Mihaylov, Peter Clark, Tushar Khot, and Ashish Sabharwal. Can a suit of armor conduct electricity? a new dataset for open book question answering, 2018

    2018

  20. [28]

    Rossi, and Thien Huu Nguyen

    Thuat Nguyen, Chien Van Nguyen, Viet Dac Lai, Hieu Man, Nghia Trung Ngo, Franck Dernoncourt, Ryan A. Rossi, and Thien Huu Nguyen. Culturax: A cleaned, enormous, and multilingual dataset for large language models in 167 languages, 2023

    2023

  21. [29]

    Thai tnhc2 books, 2024

    Wannaphong Phatthiyaphaibun. Thai tnhc2 books, 2024. URL https://doi.org/10.5281/zenodo.10783421

    2024 doi

  22. [30]

    P y T hai NLP : T hai natural language processing in P ython

    Wannaphong Phatthiyaphaibun, Korakot Chaovavanich, Charin Polpanumas, Arthit Suriyawongkul, Lalita Lowphansirikul, Pattarawat Chormai, Peerat Limkonchotiwat, Thanathip Suntorntip, and Can Udomcharoenchaikit. P y T hai NLP : T hai natural language processing in P ython. In Lili...

    2023

  23. [31]

    Ponti, Goran Glava s , Olga Majewska, Qianchu Liu, Ivan Vuli\' c , and Anna Korhonen

    Edoardo M. Ponti, Goran Glava s , Olga Majewska, Qianchu Liu, Ivan Vuli\' c , and Anna Korhonen. XCOPA: A multilingual dataset for causal commonsense reasoning. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020. URL https:/...

    2020

  24. [32]

    Winogrande: an adversarial winograd schema challenge at scale

    Keisuke Sakaguchi, Ronan Le Bras, Chandra Bhagavatula, and Yejin Choi. Winogrande: an adversarial winograd schema challenge at scale. Commun. ACM, 64 0 (9): 0 99–106, aug 2021. ISSN 0001-0782. doi:10.1145/3474381. URL https://doi.org/10.1145/3474381

    2021 doi

  25. [33]

    Thai literature corpora

    Jitkapat Sawatphol. Thai literature corpora. ://attapol.github.io/tlc.html, 2019

    2019

  26. [34]

    Glu variants improve transformer, 2020

    Noam Shazeer. Glu variants improve transformer, 2020

    2020

  27. [35]

    Yang, Bin Wu, Laurence Aitchison, Emine Yilmaz, and Aldo Lipani

    Zhengyan Shi, Adam X. Yang, Bin Wu, Laurence Aitchison, Emine Yilmaz, and Aldo Lipani. Instruction tuning with loss over instructions, 2024. URL https://arxiv.org/abs/2405.14394

    2024 arXiv

  28. [36]

    H2o-danube-1.8b technical report, 2024

    Philipp Singer, Pascal Pfeiffer, Yauhen Babakhin, Maximilian Jeblick, Nischay Dhankhar, Gabor Fodor, and Sri Satish Ambati. H2o-danube-1.8b technical report, 2024

    2024

  29. [37]

    Roformer: Enhanced transformer with rotary position embedding, 2023

    Jianlin Su, Yu Lu, Shengfeng Pan, Ahmed Murtadha, Bo Wen, and Yunfeng Liu. Roformer: Enhanced transformer with rotary position embedding, 2023

    2023

  30. [38]

    Anwer, Michael Felsberg, Tim Baldwin, Eric P

    Omkar Thawakar, Ashmal Vayani, Salman Khan, Hisham Cholakal, Rao M. Anwer, Michael Felsberg, Tim Baldwin, Eric P. Xing, and Fahad Shahbaz Khan. Mobillama: Towards accurate and lightweight fully transparent gpt, 2024

    2024

  31. [39]

    iapp\_wiki\_qa\_squad, February 2021

    Kobkrit Viriyayudhakorn and Charin Polpanumas. iapp\_wiki\_qa\_squad, February 2021. URL https://doi.org/10.5281/zenodo.4539916

    2021 doi

  32. [40]

    Xmodel-lm technical report, 2024 a

    Yichuan Wang, Yang Liu, Yu Yan, Qun Wang, Xucheng Huang, and Ling Jiang. Xmodel-lm technical report, 2024 a . URL https://arxiv.org/abs/2406.02856

    2024 arXiv

  33. [41]

    Telechat technical report, 2024 b

    Zihan Wang, Xinzhang Liu, Shixuan Liu, Yitong Yao, Yuyao Huang, Zhongjiang He, Xuelong Li, Yongxiang Li, Zhonghao Che, Zhaoxi Zhang, Yan Wang, Xin Wang, Luwen Pu, Huihan Xu, Ruiyu Fang, Yu Zhao, Jie Zhang, Xiaomeng Huang, Zhilong Lu, Jiaxin Peng, Wenjun Zheng, Shiquan Wang, Bi...

    2024

  34. [42]

    Polylm: An open source polyglot large language model, 2023

    Xiangpeng Wei, Haoran Wei, Huan Lin, Tianhao Li, Pei Zhang, Xingzhang Ren, Mei Li, Yu Wan, Zhiwei Cao, Binbin Xie, Tianxiang Hu, Shangjie Li, Binyuan Hui, Bowen Yu, Dayiheng Liu, Baosong Yang, Fei Huang, and Jun Xie. Polylm: An open source polyglot large language model, 2023. ...

    2023 arXiv

  35. [43]

    Liu, and Matt Gardner

    Johannes Welbl, Nelson F. Liu, and Matt Gardner. Crowdsourcing multiple choice science questions, 2017

    2017

  36. [44]

    Effective long-context scaling of foundation models, 2023

    Wenhan Xiong, Jingyu Liu, Igor Molybog, Hejia Zhang, Prajjwal Bhargava, Rui Hou, Louis Martin, Rashi Rungta, Karthik Abinav Sankararaman, Barlas Oguz, Madian Khabsa, Han Fang, Yashar Mehdad, Sharan Narang, Kshitiz Malik, Angela Fan, Shruti Bhosale, Sergey Edunov, Mike Lewis, S...

    2023 arXiv

  37. [45]

    mt5: A massively multilingual pre-trained text-to-text transformer, 2021

    Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, and Colin Raffel. mt5: A massively multilingual pre-trained text-to-text transformer, 2021. URL https://arxiv.org/abs/2010.11934

    2021 arXiv

  38. [46]

    Hellaswag: Can a machine really finish your sentence?, 2019

    Rowan Zellers, Ari Holtzman, Yonatan Bisk, Ali Farhadi, and Yejin Choi. Hellaswag: Can a machine really finish your sentence?, 2019

    2019

  39. [47]

    Root mean square layer normalization

    Biao Zhang and Rico Sennrich. Root mean square layer normalization. Curran Associates Inc., Red Hook, NY, USA, 2019

    2019

  40. [48]

    Map-neo: Highly capable and transparent bilingual large language model series, 2024 a

    Ge Zhang, Scott Qu, Jiaheng Liu, Chenchen Zhang, Chenghua Lin, Chou Leuang Yu, Danny Pan, Esther Cheng, Jie Liu, Qunshu Lin, Raven Yuan, Tuney Zheng, Wei Pang, Xinrun Du, Yiming Liang, Yinghao Ma, Yizhi Li, Ziyang Ma, Bill Lin, Emmanouil Benetos, Huan Yang, Junting Zhou, Kaiji...

    2024 arXiv

  41. [49]

    Tinyllama: An open-source small language model, 2024 b

    Peiyuan Zhang, Guangtao Zeng, Tianduo Wang, and Wei Lu. Tinyllama: An open-source small language model, 2024 b

    2024

  42. [50]

    Opt: Open pre-trained transformer language models

    Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen, Christopher Dewan, Mona Diab, Xian Li, Xi Victoria Lin, et al. Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068, 2022

    2022 arXiv

  43. [51]

    Patil, Naman Jain, Sheng Shen, Matei Zaharia, Ion Stoica, and Joseph E

    Tianjun Zhang, Shishir G. Patil, Naman Jain, Sheng Shen, Matei Zaharia, Ion Stoica, and Joseph E. Gonzalez. Raft: Adapting language model to domain specific rag, 2024 c . URL https://arxiv.org/abs/2403.10131

    2024 arXiv

  44. [52]

    Instruction-following evaluation for large language models, 2023

    Jeffrey Zhou, Tianjian Lu, Swaroop Mishra, Siddhartha Brahma, Sujoy Basu, Yi Luan, Denny Zhou, and Le Hou. Instruction-following evaluation for large language models, 2023. URL https://arxiv.org/abs/2311.07911

    2023 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.