Pith. sign in

REVIEW 3 major objections 5 minor 57 references

A Comparative Study of Task Adaptation Techniques of Large Language Models for Identifying Sustainable Development Goals

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A fine-tuned 13-billion-parameter LLaMa-2 model classifies UN Sustainable Development Goal texts at 92.4% F1, slightly beating fine-tuned GPT-3.5; smaller models can match larger ones when prompts and fine-tuning are chosen well.

desk verdict Useful wide comparison of LLMs for SDG classification, with a real benchmark flaw in the 'Other' class that should be fixed before the absolute numbers are trusted. read the letter →

arxiv 2506.15208 v1 pith:JLOKSRE2 submitted 2025-06-18 cs.CL cs.AI

Andrea Cadeddu , Alessandro Chessa , Vincenzo De Leo , Gianni Fenu , Enrico Motta , Francesco Osborne , Diego Reforgiato Recupero , Angelo Salatino
show 1 more author
Luca Secchi
This is my paper · ORCID
classification cs.CLcs.AI
keywords largelanguagemodelstextclassificationsustainabledevelopmentgoalsfew-shotlearningzero-shotfine-tuningpromptengineeringmodelquantization
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 asks whether small open-source language models, guided by well-chosen prompts and fine-tuning, can match or beat much larger proprietary models at assigning short texts to the UN Sustainable Development Goals. It builds a balanced 17-class benchmark from volunteer-validated excerpts in the OSDG Community Dataset and evaluates 18 models under zero-shot learning, few-shot learning, and fine-tuning. Its headline finding is that fine-tuned LLaMa-2 13B reaches 92.4% F1, ahead of fine-tuned GPT-3.5-turbo at 91.4%, while few-shot learning with semantically similar exemplars adds roughly 20 F1 points over random selection. The paper concludes that task adaptation, not raw parameter count, drives performance, making accurate SDG classification feasible with relatively small open-weight models.

What carries the argument

The benchmark itself is the load-bearing object: 6,800 excerpts, 400 per class, covering SDGs 1–16 (from the OSDG Community Dataset, filtered to high volunteer agreement) plus an 'Other' class built from 400 online news articles, split 4,760/1,020/1,020 for training, validation, and test. Around it, the paper varies only the adaptation machinery: zero-shot prompts; few-shot prompts with 3, 5, or 7 exemplars chosen randomly, by cosine similarity, or by cosine similarity with a distinct-class constraint; and fine-tuning via LoRA (low-rank adaptation) or QLoRA (quantized LoRA) for 4-bit models. The mechanism that carries the argument is the interaction between selection strategy and model: semantic-similarity exemplars give the largest few-shot gains, and LoRA fine-tuning is what pushes a 13B open model past a larger proprietary one.

What would settle it

Reserve a test subset where 'Other' texts are drawn from the same UN report and policy-document sources as the SDG classes (manually verified to mention none of the 16 goals); if fine-tuned LLaMa-2 13B's F1 on that subset falls well below its overall 92.4%, the benchmark rewards a genre shortcut. A cleaner test would be a two-by-two design crossing document genre with SDG relevance and measuring whether models classify relevance within each genre.

Watch

Extended reading notes

Core claim

The paper's central claim is that for this task the adaptation technique matters more than model scale. In zero-shot settings, the GPT models lead (GPT-3.5 at 81.1%, GPT-4 at 80.3%), but fine-tuning reverses the order: LLaMa-2 13B reaches 92.4%, GPT-3.5-turbo 91.4%, and Flan-T5-XXL 90.7%. In few-shot settings, selecting exemplars by cosine semantic similarity outperforms random selection by 19.3–21.5 F1 points and outperforms similarity-with-different-classes, with LLaMa-3 8B peaking at 3 examples, GPT-4 at 5, and Mixtral-8x7B at 7. Fine-tuning improves every model by an average of 25.7 percentage points over zero-shot, and 4-bit quantized variants such as LLaMa-2 13B (85.7%) and Zephyr 7B (86.4%) remain competitive.

Load-bearing premise

The benchmark's 'Other' class comes from online news articles while the 16 SDG classes come from UN reports and policy documents, so the models may be separating text by genre rather than by SDG relevance.

Editorial extensions

If this is right

  • Fine-tuned open-weight models such as LLaMa-2 13B can serve as accurate SDG classifiers, reaching about 92% F1 on a balanced 17-class benchmark.
  • When fine-tuning is unavailable, few-shot prompts should be built by retrieving semantically similar training excerpts; random exemplars give essentially zero benefit over zero-shot.
  • Quantized 4-bit models (LLaMa-2 13B 4-bit at 85.7%, Zephyr 7B 4-bit at 86.4%) allow near-competitive SDG classification without large GPU clusters.
  • Parameter count does not determine rank: BERT at 110M parameters beats LLaMa-3 8B when both are fine-tuned, and Flan-T5-base at 220M reaches 90.5%.
  • The optimal number of few-shot exemplars is model-specific, so deployment should tune the example count rather than assume more examples are always better.

Reading between the lines

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

  • If the 'Other' class were drawn from UN-style reports and policy documents instead of online news, classification could get harder; part of the reported accuracy may come from distinguishing news-like from report-like text rather than SDG relevance.
  • The semantic-similarity result generalizes as a practical recipe: for multi-class text tasks, retrieval-based exemplar selection should be the default few-shot setup, not random sampling.
  • Because GPT-3.5 edged GPT-4 in zero-shot, the paper implies that prompt configuration matters as much as model generation; testing systematic prompt search across newer models would clarify when scale helps.
  • The benchmark's single-label design leaves multi-label SDG overlap unexamined, even though real documents often address several goals at once; a natural extension is to evaluate on texts with multiple valid labels.
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

3 major / 5 minor

Summary. The paper presents a comparative evaluation of 18 language models for single-label multi-class classification of text into 16 SDG classes plus an 'Other' class. Using a benchmark extracted from the OSDG Community Dataset (October 2023), with balanced train/validation/test splits (280/60/60 examples per class), the authors evaluate zero-shot, few-shot (3/5/7 examples with three selection strategies: random, semantic similarity, and semantic similarity across different classes), and fine-tuning. They report macro-averaged precision, recall, accuracy, and F1. Main empirical findings: fine-tuned LLaMa-2-13B achieves F1 92.4%, GPT-3.5-turbo FT 91.4%, Flan-T5-XXL 90.9%; semantic similarity-based example selection outperforms random and diverse-class selection; quantized models remain competitive. The paper concludes that smaller open models can match or exceed larger proprietary models when fine-tuned on high-quality data.

Significance. If the benchmark is valid, this study provides practically useful, falsifiable evidence about model and adaptation-technique choices for SDG text classification. Its strengths include the use of a public volunteer-labelled dataset, public prompt templates, coverage of 18 models across three adaptation families, explicit hyperparameters for fine-tuning, and consistent internal tables. The main caveat is that the negative class ('Other') is constructed from a different document distribution than the 16 SDG classes, which must be addressed before the F1 comparisons can be interpreted as measuring SDG-relevance discrimination rather than genre or source discrimination.

major comments (3)
  1. [Section III-A] The construction of the 'Other' class is a load-bearing issue for the benchmark claim. The 16 SDG classes are drawn from UN reports, policy documents, and publication abstracts, while the 400 'Other' texts are sourced from online news websites and manually verified as non-SDG. With a balanced 17-class test set, macro-F1 weights 'Other' equally with each SDG class, and a classifier can separate 'Other' from the positive classes using source/genre cues rather than SDG relevance. The paper reports no per-class F1 and no source-stratified analysis that would rule out this shortcut. I recommend reporting per-class F1 (especially for 'Other'), and ideally adding a matched-distribution negative control, e.g., non-SDG excerpts from the same UN/policy/publication document sources, or at least an explicit quantitative analysis of the genre cue. Without this, the central comparison is not yet established as a measure of SDG-identification performance.
  2. [Section V / Tables 3-7] All reported results appear to come from single runs, with no confidence intervals, standard deviations, or significance tests. Topline differences used for ranking (LLaMa-2-13B 92.4 vs GPT-3.5 91.4 vs Flan-T5-XXL 90.9) are within one or two points and could easily be within run-to-run or test-set noise. I recommend adding repeated runs with multiple seeds for open models and reporting mean and standard deviation; for API models, statistical tests over the test set (e.g., McNemar's test or bootstrap confidence intervals) would substantially strengthen the ranking claims.
  3. [Section IV] There is a direct contradiction about which models were evaluated in zero-shot. The text states: 'Due to resource limitations, we conducted ZSL experiments on all models except LLaMa-2-70b-chat-hf', yet Table 2 and Table 3 both include Llama-2-70b-chat-hf with ZSL results. This is almost certainly a typo (the intended exception is probably fine-tuning), but it must be corrected, as the current wording undermines the reproducibility of the experimental protocol.
minor comments (5)
  1. [Section V] The sentence reporting 'an average increase in the F1 score of 25.7% points compared to the ZSL' is unclear because not all fine-tuned models have a corresponding ZSL result; please specify exactly which subset of models the average is over.
  2. [Table 8] The caption reads 'ZSL focused on the SEM. SIMILARITY method', but the FS-SS columns are few-shot results; the caption should say 'FSL with semantic similarity'.
  3. [References] Reference [2] appears to share the same URL as reference [1]; please verify and correct the URL for reference [2].
  4. [Section IV] There is a typo: 'Zephir' in the fine-tuning hyperparameter description should be 'Zephyr'.
  5. [Section III-B] The text describes BERT as one of the '18 language models' but BERT is an encoder-only model and not typically considered an LLM; please clarify the terminology or describe BERT as a baseline model.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the study is an empirical benchmark evaluation, not a derivation, and its headline claims are read directly from measured F1 scores.

full rationale

The paper performs an empirical comparison of 18 LLMs on a fixed SDG text-classification benchmark derived from the external OSDG Community Dataset. There is no derivation chain in which an input quantity is re-labeled as a prediction: fine-tuning and prompting are evaluated on a held-out test set, and the reported F1 values are measured outcomes rather than fitted constants renamed as results. The only author-added component is the 'Other' class, constructed by manually selecting 400 online-news texts judged not to refer to any SDG (Section III-A). This is a dataset-construction choice, not a circular step: the models are not given the manual judgments at inference time, and the manual screen does not define the metric used to rank models. The possible genre/source mismatch between the 'Other' texts and the UN-report/policy-document SDG texts is a validity concern about what the benchmark measures, not a circularity in the paper's reasoning. Self-citations appear in the related-work discussion (e.g., references [24], [27], [28], [34]) and support contextual statements about prior applications of LLMs and knowledge injection; none of these citations is load-bearing for the central empirical claim that fine-tuned LLaMa-2 13B reaches 92.4% F1. The central claim is self-contained against an external benchmark and does not reduce to its inputs by definition.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the quality of the constructed benchmark: the OSDG-CD volunteer labels, the hand-picked filtering thresholds, the balanced sample sizes, and the manual construction of the 'Other' class. No new entities or fitted mathematical constants are introduced.

free parameters (3)
  • Agreement threshold = 0.75
    Texts included only if volunteer agreement >= 0.75 and positive votes > 2x negative votes; this hand-picked threshold shapes the gold-standard benchmark and therefore all reported F1 scores.
  • Per-class sample size = 400
    Benchmark balanced with 400 texts per class (16 SDG classes plus 400 'Other' news texts). Choice affects class balance and test-set size (60 per class).
  • FSL example counts = 3, 5, 7
    Three hand-chosen values for number of in-context examples; best results per model vary with S, so conclusions depend on this grid.
assumptions (3)
  • domain assumption OSDG Community Dataset volunteer labels provide reliable ground truth for SDG relevance
    The benchmark and all metrics are built on these labels after filtering; if annotations are noisy, comparisons are affected.
  • ad hoc to paper Manual verification that 400 news texts are not SDG-related is correct
    No inter-annotator agreement or external validation is reported for this 'Other' class construction.
  • domain assumption Macro-averaged F1 is the appropriate aggregation metric
    Used because classes are equally important, but it treats the 'Other' class exactly like each SDG despite its different provenance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Comparative Study of Task Adaptation Techniques of Large Language Models for Identifying Sustainable Development Goals." pith.science (2026). https://pith.science/paper/JLOKSRE2

@misc{pith2026250615208,
  author       = {Pith},
  title        = {Pith review of: A Comparative Study of Task Adaptation Techniques of Large Language Models for Identifying Sustainable Development Goals},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JLOKSRE2}},
  note         = {Machine review of arXiv:2506.15208}
}
read the original abstract

In 2012, the United Nations introduced 17 Sustainable Development Goals (SDGs) aimed at creating a more sustainable and improved future by 2030. However, tracking progress toward these goals is difficult because of the extensive scale and complexity of the data involved. Text classification models have become vital tools in this area, automating the analysis of vast amounts of text from a variety of sources. Additionally, large language models (LLMs) have recently proven indispensable for many natural language processing tasks, including text classification, thanks to their ability to recognize complex linguistic patterns and semantics. This study analyzes various proprietary and open-source LLMs for a single-label, multi-class text classification task focused on the SDGs. Then, it also evaluates the effectiveness of task adaptation techniques (i.e., in-context learning approaches), namely Zero-Shot and Few-Shot Learning, as well as Fine-Tuning within this domain. The results reveal that smaller models, when optimized through prompt engineering, can perform on par with larger models like OpenAI's GPT (Generative Pre-trained Transformer).

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 40 canonical work pages

  1. [1]

    Leadership Council Sustainable Development Solutions Network, ‘‘Indicators and a Monitoring Framework for the Sustainable Development Goals’’, Sep. 2015. [Online]. Available: https://sdgs.un.org/publications/ indicators-and-monitoring-framework-sustainable-development-goals-17958

    work page 2015

  2. [2]

    Sustainable Development Solutions Network, ‘‘Data for Development - A Needs Assessment for SDG Monitoring and Statistical Capacity Develop- ment’’, Apr. 2015. [Online]. Available: https://sdgs.un.org/publications/ indicators-and-monitoring-framework-sustainable-development-goals-17958

    work page 2015

  3. [3]

    Hirai and F

    T. Hirai and F. Comim, ‘‘Measuring the sustainable development goals: A poset analysis’’,Ecological Indicators, vol. 145, p. 109605, 2022

    work page 2022

  4. [4]

    Ziegler et al., ‘‘How to Measure Sustainability? An Open-Data Ap- proach’’,Sustainability, vol

    D. Ziegler et al., ‘‘How to Measure Sustainability? An Open-Data Ap- proach’’,Sustainability, vol. 15, no. 4, pp. 3203–, Feb. 2023

    work page 2023

  5. [5]

    M. Mishra et al., ‘‘A bibliometric analysis of sustainable develop- ment goals (SDGs): a review of progress, challenges, and opportuni- ties’’,Environment, Development and Sustainability, pp. 1–43, 2023, doi: 10.1007/s10668-023-03225-w

    doi:10.1007/s10668-023-03225-w 2023

  6. [6]

    La Fleur, ‘‘Using large language models to help train machine learning SDG classifiers’’, UN DESA, Working Paper 180, Nov

    M. La Fleur, ‘‘Using large language models to help train machine learning SDG classifiers’’, UN DESA, Working Paper 180, Nov. 2023

    work page 2023

  7. [7]

    Nilashi, G

    M. Nilashi, G. Boon Ooi, G. Tan, B. Lin, and R. Abumalloh, ‘‘Critical Data Challenges in Measuring the Performance of Sustainable Development Goals: Solutions and the Role of Big-Data Analytics’’,Harvard Data Science Review, vol. 5, no. 3, 2023

    work page 2023

  8. [8]

    J. E. Guisiano, R. Chiky, and J. De Mello, ‘‘SDG-Meter: A Deep Learning Based Tool for Automatic Text Classification of the Sustainable Devel- opment Goals’’, inIntelligent Information and Database Systems, Cham: Springer International Publishing, 2022, pp. 259–271

    work page 2022

Show all 57 references
  1. [9]

    T. Matsui et al., ‘‘A natural language processing model for supporting sustainable development goals: translating semantics, visualizing nexus, and connecting stakeholders’’,Sustainability Science, vol. 17, no. 3, pp. 969–985, May 2022, doi: 10.1007/s11625-022-01093-3

    2022 doi

  2. [10]

    Guisiano and R

    J. Guisiano and R. Chiky, ‘‘Automatic classification of multilabel texts related to Sustainable Development Goals (SDGs)’’, inTECHENV EGC2021, Montpellier, France, Jan. 2021. [Online]. Available: https://hal. archives-ouvertes.fr/hal-03154261

    2021

  3. [11]

    Heras-Saizarbitoria, L

    I. Heras-Saizarbitoria, L. Urbieta, and O. Boiral, ‘‘Organizations’ engage- ment with sustainable development goals: From cherry-picking to SDG- washing?’’,Corporate Social Responsibility and Environmental Manage- ment, vol. 29, no. 2, pp. 316–328, 2022

    2022

  4. [12]

    Pukelis, N

    L. Pukelis, N. Bautista-Puig, M. Skrynik, and V . Stanciauskas, ‘‘OSDG - Open-Source Approach to Classify Text Data by UN Sustainable Develop- ment Goals (SDGs)’’,CoRR, abs/2005.14569, 2020. [Online]. Available: https://arxiv.org/abs/2005.14569

    2005 arXiv

  5. [13]

    V aswani et al., ‘‘Attention is all you need’’, inProc

    A. V aswani et al., ‘‘Attention is all you need’’, inProc. 31st Intl. Conf. on Neural Information Processing Systems, 2017, pp. 6000–6010

    2017

  6. [14]

    Devlin, M.-W

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, ‘‘BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding’’, inProc. 2019 Conf. of the North American Chapter of the Association for Compu- tational Linguistics: Human Language Technologies, vol. 1, Min...

    2019

  7. [15]

    Brown et al., ‘‘Language Models are Few-Shot Learners’’, inAdvances in Neural Information Processing Systems, vol

    T. Brown et al., ‘‘Language Models are Few-Shot Learners’’, inAdvances in Neural Information Processing Systems, vol. 33, 2020, pp. 1877–1901

    2020

  8. [16]

    Joshi, L

    A. Joshi, L. G. Gonzalez Morales, S. Klarman, A. Stellato, A. Helton, C. S. Lovell, and A. Haczek, ‘‘A Knowledge Organization System for the United Nations Sustainable Development Goals’’, inEighteenth Extended Semantic Web Conference - Resources Track, 2021. [Online]. Availab...

    2021

  9. [17]

    Salatino, T

    A. Salatino, T. Aggarwal, A. Mannocci, F. Osborne, and E. Motta, ‘‘A survey on knowledge organization systems of research fields: Resources and challenges’’,arXiv preprint arXiv:2409.04432, 2024

    2024 arXiv

  10. [18]

    Sovrano, M

    F. Sovrano, M. Palmirani, and F. Vitali, ‘‘Deep learning based multi-label text classification of UNGA resolutions’’, inProc. 13th Intl. Conf. on Theory and Practice of Electronic Governance, Athens, Greece, 2020, pp. 686–695, doi: 10.1145/3428502.3428604

    2020

  11. [19]

    Schmidt and M

    F. Schmidt and M. V anderfeesten, ‘‘Evaluation on accuracy of mapping science to the United Nations’ Sustainable Development Goals (SDGs) of the Aurora SDG queries’’, Zenodo, v1.0.2, Jun. 2021, doi: 10.5281/zen- odo.4964606

    2021 doi

  12. [20]

    Bolanos, A

    F. Bolanos, A. Salatino, F. Osborne, and E. Motta, ‘‘Artificial intelligence for literature reviews: Opportunities and challenges’’,Artificial Intelli- gence Review, vol. 57, no. 259, 2024

    2024

  13. [21]

    C. W. Kosonocky, C. O. Wilke, E. M. Marcotte, and A. D. Ellington, ‘‘Min- ing patents with large language models elucidates the chemical function landscape’’,Digital Discovery, vol. 3, no. 6, pp. 1150–1159, 2024, Royal Society of Chemistry

    2024

  14. [22]

    J. A. Omiye, H. Gui, S. J. Rezaei, J. Zou, and R. Daneshjou, ‘‘Large lan- guage models in medicine: the potentials and pitfalls: a narrative review’’, Annals of internal medicine, vol. 177, no. 2, pp. 210–220, 2024, American College of Physicians

    2024

  15. [23]

    Y ang, T

    K. Y ang, T. Zhang, Z. Kuang, Q. Xie, J. Huang, and S. Ananiadou, ‘‘MentaLLaMA: interpretable mental health analysis on social media with large language models’’, inProc. ACM Web Conference 2024, 2024, pp. 4489–4500

    2024

  16. [24]

    Motta, F

    E. Motta, F. Osborne, M. M. L. Pulici, A. Salatino, and I. Naja, ‘‘Capturing the Viewpoint Dynamics in the News Domain’’, inIntl. Conf. on Knowl- edge Engineering and Knowledge Management, 2024, pp. 18–34, Springer

    2024

  17. [25]

    Cascini, A

    G. Cascini, A. Fantechi, and E. Spinicci, ‘‘Natural language processing of patents and technical documentation’’, inIntl. Workshop on Document Analysis Systems, 2004, pp. 508–520, Springer

    2004

  18. [26]

    Savelka and K

    J. Savelka and K. D. Ashley, ‘‘The unreasonable effectiveness of large language models in zero-shot semantic annotation of legal texts’’,Frontiers in Artificial Intelligence, vol. 6, p. 1279794, 2023, Frontiers Media SA

    2023

  19. [27]

    C. Peng, F. Xia, M. Naseriparsa, and F. Osborne, ‘‘Knowledge graphs: opportunities and challenges’’,Artificial Intelligence Review, pp. 1–32, 2023, Springer

    2023

  20. [28]

    Cadeddu, A

    A. Cadeddu, A. Chessa, V . De Leo, G. Fenu, E. Motta, F. Osborne, D. R. Recupero, A. Salatino, and L. Secchi, ‘‘Optimizing Tourism Accom- modation Offers by Integrating Language Models and Knowledge Graph Technologies’’,Information, vol. 15, no. 7, p. 398, 2024, MDPI

    2024

  21. [29]

    Y . Li, S. Wang, H. Ding, and H. Chen, ‘‘Large language models in finance: A survey’’, inProc. F ourth ACM Int. Conf. on AI in Finance, 2023, pp. 374–382

    2023

  22. [30]

    Zhang, M

    Y . Zhang, M. Wang, C. Ren, Q. Li, P . Tiwari, B. Wang, and J. Qin, ‘‘Pushing the limit of LLM capacity for text classification’’,arXiv preprint arXiv:2402.07470, 2024

    2024 arXiv

  23. [31]

    M. S. U. Miah, M. M. Kabir, T. B. Sarwar, M. Safran, S. Alfarhood, and M. F. Mridha, ‘‘A multimodal approach to cross-lingual sentiment analysis with ensemble of transformer and LLM’’,Scientific Reports, vol. 14, no. 1, p. 9603, 2024, Nature Publishing Group UK London

    2024

  24. [32]

    P . Wang, K. He, Y . Wang, X. Song, Y . Mou, J. Wang, Y . Xian, X. Cai, and W. Xu, ‘‘Beyond the known: Investigating llms performance on out-of-domain intent detection’’,arXiv preprint arXiv:2402.17256, 2024

    2024 arXiv

  25. [33]

    L. Tang, P . Laban, and G. Durrett, ‘‘Minicheck: Efficient fact-checking of llms on grounding documents’’,arXiv preprint arXiv:2404.10774, 2024

    2024 arXiv

  26. [34]

    Cadeddu, A

    A. Cadeddu, A. Chessa, V . De Leo, G. Fenu, E. Motta, F. Osborne, D. R. Recupero, A. Salatino, and L. Secchi, ‘‘A comparative analysis of knowledge injection strategies for large language models in the scholarly domain’’,Engineering Applications of Artificial Intelligence, vol...

    2024

  27. [35]

    Radford and K

    A. Radford and K. Narasimhan, ‘‘Improving Language Understand- ing by Generative Pre-Training’’, 2018. [Online]. Available: https://api. semanticscholar.org/CorpusID:49313245

    2018

  28. [36]

    Fields, K

    J. Fields, K. Chovanec, and P . Madiraju, ‘‘A Survey of Text Classification With Transformers: How Wide? How Large? How Long? How Accurate? How Expensive? How Safe?’’,IEEE Access, vol. 12, pp. 6518–6531, 2024. VOLUME 11, 2023 13 Authoret al.: Preparation of Papers for IEEE TRA...

    2024

  29. [37]

    S. M. Jain,Introduction to Transformers for NLP: With the Hugging Face Library and Models to Solve Problems. Berkeley, CA, USA: Apress, 2022

    2022

  30. [38]

    Q. Li, H. Peng, J. Li, C. Xia, R. Y ang, L. Sun, P . S. Y u, and L. He, ‘‘A Survey on Text Classification: From Traditional to Deep Learning’’,ACM Trans. Intell. Syst. Technol., vol. 13, no. 2, p. 31, Apr. 2022, doi: 10.1145/3495162

    2022 doi

  31. [39]

    Gasparetto, M

    A. Gasparetto, M. Marcuzzo, A. Zangari, and A. Albarelli, ‘‘A Survey on Text Classification Algorithms: From Text to Predictions’’,Information, vol. 13, no. 2, p. 83, 2022, doi: 10.3390/info13020083

    2022 doi

  32. [40]

    Y ang, H

    J. Y ang, H. Jin, R. Tang, X. Han, Q. Feng, H. Jiang, S. Zhong, B. Yin, and X. Hu, ‘‘Harnessing the Power of LLMs in Practice: A Survey on ChatGPT and Beyond’’,ACM Trans. Knowl. Discov. Data, 2024, doi: 10.1145/3649506. [Online]. Available: https://doi.org/10.1145/3649506

    2024 doi

  33. [41]

    Bisong,Building Machine Learning and Deep Learning Models on Google Cloud Platform: A Comprehensive Guide for Beginners

    E. Bisong,Building Machine Learning and Deep Learning Models on Google Cloud Platform: A Comprehensive Guide for Beginners. Berkeley, CA, USA: Apress, 2019

    2019

  34. [42]

    Dettmers, A

    T. Dettmers, A. Pagnoni, A. Holtzman, and L. Zettlemoyer, ‘‘QLoRA: Effi- cient Finetuning of Quantized LLMs’’, inAdvances in Neural Information Processing Systems, vol. 36, 2023, pp. 10088–10115

    2023

  35. [43]

    E. J. Hu et al., ‘‘LoRA: Low-Rank Adaptation of Large Language Models’’, inIntl. Conf. on Learning Representations, 2022. [Online]. Available: https://openreview.net/forum?id=nZeVKeeFYf9

    2022

  36. [44]

    Achiam et al., ‘‘GPT-4 Technical Report’’,arXiv preprint arXiv:2303.08774, 2024

    J. Achiam et al., ‘‘GPT-4 Technical Report’’,arXiv preprint arXiv:2303.08774, 2024

    2024 arXiv

  37. [45]

    Touvron et al., ‘‘Llama 2: Open Foundation and Fine-Tuned Chat Models’’,arXiv preprint arXiv:2307.09288, 2023

    H. Touvron et al., ‘‘Llama 2: Open Foundation and Fine-Tuned Chat Models’’,arXiv preprint arXiv:2307.09288, 2023

    2023 arXiv

  38. [46]

    Ainslie, J

    J. Ainslie, J. Lee-Thorp, M. de Jong, Y . Zemlyanskiy, F. Lebron, and S. Sanghai, ‘‘GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints’’, inProc. 2023 Conf. on Empirical Methods in Natural Language Processing, Singapore, 2023, pp. 4895–4901

    2023

  39. [47]

    H. W. Chung et al., ‘‘Scaling Instruction-Finetuned Language Models’’, arXiv preprint arXiv:2210.11416, 2022

    2022 arXiv

  40. [48]

    Raffel et al., ‘‘Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer’’,J

    C. Raffel et al., ‘‘Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer’’,J. Mach. Learn. Res., vol. 21, no. 140, pp. 1– 67, 2020

    2020

  41. [49]

    A. Q. Jiang et al., ‘‘Mistral 7B’’,arXiv preprint arXiv:2310.06825, 2023

    2023 arXiv

  42. [50]

    A. Q. Jiang et al., ‘‘Mixtral of Experts’’,arXiv preprint arXiv:2401.04088, 2024

    2024 arXiv

  43. [51]

    Tunstall, E

    L. Tunstall, E. Beeching, N. Lambert, N. Rajani, K. Rasul, Y . Belkada, S. Huang, L. von Werra, C. Fourrier, N. Habib, N. Sarrazin, O. Sanseviero, A. M. Rush, and T. Wolf, ‘‘Zephyr: Direct Distillation of LM Alignment’’, arXiv preprint arXiv:2310.16944, 2023

    arXiv

  44. [52]

    Abdin et al., ‘‘Phi-3 Technical Report: A Highly Capable Language Model Locally on Y our Phone’’,arXiv preprint arXiv:2404.14219, 2024

    M. Abdin et al., ‘‘Phi-3 Technical Report: A Highly Capable Language Model Locally on Y our Phone’’,arXiv preprint arXiv:2404.14219, 2024

    2024 arXiv

  45. [53]

    J. Wei, M. Bosma, V . Zhao, K. Guu, A. W. Y u, B. Lester, N. Du, A. M. Dai, and Q. V . Le, ‘‘Finetuned Language Models are Zero-Shot Learners’’, in Intl. Conf. on Learning Representations, 2022. [Online]. Available: https: //openreview.net/forum?id=gEZrGCozdqR

    2022

  46. [54]

    Edwards and J

    A. Edwards and J. Camacho-Collados, ‘‘Language Models for Text Classification: Is In-Context Learning Enough?’’,arXiv preprint arXiv:2403.17661, 2024

    2024 arXiv

  47. [55]

    H. Y u, Z. Y ang, K. Pelrine, J. F. Godbout, and R. Rabbany, ‘‘Open, Closed, or Small Language Models for Text Classification?’’,arXiv preprint arXiv:2308.10092, 2023

    2023 arXiv

  48. [56]

    G. O. Y oung, ‘‘Synthetic structure of industrial plastics’’, inPlastics,2 nd ed., vol. 3, J. Peters, Ed. New Y ork, NY , USA: McGraw-Hill, 1964, pp. 15–64. ANDREA CADEDDUcurrently works as a data scientist at Linkalab, a private Computational Lab- oratory on Complex Systems. ...

    1964

  49. [2007]

    of the Knowledge Media Institute (KMi) at the Open University in UK. Prof. Motta has a Laurea in Computer Science from the Univer- sity of Pisa in Italy and a PhD in Artificial In- telligence from the Open University. His research spans a variety of aspects at the intersection...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.