Pith. sign in

REVIEW 4 major objections 5 minor 52 references

SusGen-GPT: A Data-Centric LLM for Financial NLP and Sustainability Report Generation

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A 7-8B parameter model fine-tuned on a balanced 30K instruction dataset matches GPT-4's average performance on eight financial and ESG NLP tasks, including generating TCFD-format sustainability reports, despite GPT-4's 1,700B parameters.

desk verdict The near-GPT-4 claim collapses because the paper's own Table 3 shows the evaluation test sets were in the training data. read the letter →

arxiv 2412.10906 v1 pith:6YJ3EMQM submitted 2024-12-14 cs.CL cs.CEcs.LGq-fin.CP

Qilong Wu , Xiaoneng Xiang , Hejia Huang , Xuan Wang , Yeo Wei Jie , Ranjan Satapathy , Ricardo Shirota Filho , Bharadwaj Veeravalli This is my paper · ORCID
classification cs.CLcs.CEcs.LGq-fin.CP
keywords financialNLPESGsustainabilityreportgenerationTCFD-BenchSusGen-30KQLoRAfine-tuningretrieval-augmentedlargelanguagemodels
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 claims that a compact, open-source language model fine-tuned on a carefully curated instruction dataset can reach near-GPT-4 performance across financial and ESG natural language processing tasks. The authors introduce SusGen-30K, a category-balanced dataset covering seven financial tasks plus ESG question answering, and TCFD-Bench, a benchmark for evaluating sustainability report generation. Their trained suite of 7-8B parameter models, SusGen-GPT, attains an average score only 0.02 below GPT-4 on eight tasks while using roughly two hundred times fewer parameters. If true, this means that a data-centric fine-tuning approach can substitute for massive model scale in specialized financial and sustainability applications.

What carries the argument

SusGen-30K is a category-balanced instruction dataset built from public financial datasets and TCFD-format annual reports, processed through a pipeline of translation, reformatting, anonymization, instruction augmentation, and synthetic data generation. TCFD-Bench is a benchmark constructed from 14 ESG reports in TCFD format, with contexts generated by GPT-4o and question-answer pairs extracted manually and diversified with Mistral 7B. QLoRA fine-tuning adapts base models (Mistral-v0.3-7B and LLaMA-3-8B) to SusGen-30K, and a RAG module retrieves relevant annual-report chunks for report generation. Together these components carry the argument that a small model plus good data can rival a much larger model.

What would settle it

Check the overlap between TCFD-Bench and the TCFD_QA samples in SusGen-30K: if any company report, question, or reference answer appears in both, then the benchmark is contaminated and the reported report-generation scores do not demonstrate generalization. A concrete test is to compute the string overlap or report-identity overlap between the two sets; a nonzero overlap would falsify the off-the-shelf evaluation claim.

Watch

Extended reading notes

Core claim

The central discovery is that data curation and task balancing can close most of the gap between small open models and frontier proprietary models in finance and ESG. SusGen-GPT, fine-tuned with QLoRA on SusGen-30K, outperforms all open-source baselines and trails GPT-4 by only 2 points on average across six adapted tasks and two off-the-shelf tasks (text summarization and sustainability report generation). On the proposed TCFD-Bench, SusGen-GPT beats ChatReport on ROUGE-L, BERTScore, and METEOR, though ChatReport leads on BLEU-1. The paper presents this as evidence that a data-centric approach delivers efficient, high-performance specialized LLMs.

Load-bearing premise

The claim that sustainability report generation is an off-the-shelf, non-adapted task is contradicted by Table 3, which lists TCFD_QA as part of the ESG-QA training data in SusGen-30K; if TCFD-Bench is drawn from the same 14 TCFD reports that produced TCFD_QA, the report-generation evaluation is not held out and the measured performance reflects memorization rather than generalization.

Editorial extensions

If this is right

  • If the central claim holds, organizations can fine-tune 7-8B open models to near-frontier performance on financial and ESG tasks with modest GPU resources, avoiding dependence on closed APIs.
  • TCFD-Bench gives the community a standardized way to evaluate sustainability report generation, so future models can be compared directly on the same reports and metrics.
  • The data-centric recipe of category balancing, instruction augmentation, and synthetic data generation can be transferred to other specialized, low-resource domains.
  • The RAG-integrated SusGen system suggests a practical architecture for grounding report generation in unstructured annual reports, reducing hallucination risk.

Reading between the lines

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

  • The 'off-the-shelf' distinction is weakened by the TCFD_QA overlap; a clean evaluation would require a report-generation test set built from reports never seen during training, and the current numbers should be read with that caveat.
  • The average 0.02 gap to GPT-4 aggregates heterogeneous metrics across eight tasks, hiding large task-level deficits (e.g., GPT-4 scores 0.83 EntityF1 on NER vs. 0.35 for SusGen-GPT); parity is not uniform.
  • A more direct test of the paper's headline claim would be to evaluate GPT-4 on TCFD-Bench; currently the report-generation comparison uses ChatReport, a weaker baseline, so the claim that the model matches GPT-4 on report generation is not directly demonstrated.
  • The RAG component's contribution could be isolated by ablating retrieval; without that ablation, it is unclear how much of the TCFD-Bench score comes from retrieval versus the fine-tuned generator.
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.

Referee Report

4 major / 5 minor

Summary. The paper introduces SusGen-30K, a category-balanced instruction dataset for financial and ESG NLP, and TCFD-Bench, a benchmark for TCFD-style sustainability report generation. The authors fine-tune Mistral-7B and Llama-3-8B variants with QLoRA and report that their SusGen-GPT models outperform open baselines and trail GPT-4 by only about 0.02 in an aggregate score across six adapted and two off-the-shelf tasks. The main experimental evidence is Table 1 (FinBen-derived financial tasks) and Table 2 (TCFD-Bench vs. ChatReport).

Significance. Data-centric fine-tuning of small open models for finance and ESG is a worthwhile goal, and the paper's dataset construction pipeline and proposed benchmark address a real gap. If the comparisons were valid, a 7-8B model within 2% of GPT-4 would be practically significant. However, the headline result is not currently supported because the training set appears to include the evaluation test splits, and several comparison choices (imported GPT-4 numbers, zero-shot baselines, unweighted metric averaging) make the claimed gap difficult to interpret.

major comments (4)
  1. [Table 3 and Table 1] Table 3, rows for flare-fomc, flare-tatqa, flare-edtsum-test, flare-causal20-sc-test, flare-finer-ord, flare-mlesg-en, and flare-fiqasa: the Final counts equal the Test counts (e.g., flare-fomc Test 496 Final 496; flare-tatqa Test 1,668 Final 1,668; flare-edtsum-test Test 2,000 Final 2,000; flare-causal20-sc-test Test 8,628 Final 8,628; flare-finer-ord Test 1,075 Final 1,075; flare-mlesg-en Test 300 Final 300; flare-fiqasa Test 235 Final 235). These are the same datasets evaluated in Table 1, so the reported SusGen-GPT scores on those tasks reflect reproduction of training examples rather than generalization to held-out data. This invalidates the central claim in the Abstract and Section 4.3 that SusGen-GPT is only 0.02 below GPT-4.
  2. [Section 4.2 and Table 3] Section 4.2 states that Text Summarization and Sustainability Report Generation are non-adapted tasks because the training set 'does not explicitly include them.' This is contradicted by Table 3, which lists flare-edtsum-test under SUM with Final 2,000 and TCFD_QA under ESG-QA with Final 1,669. Additionally, TCFD-Bench is constructed from 14 TCFD-format ESG reports, and TCFD_QA comprises question-answer pairs extracted from that type of report; if the TCFD-Bench test samples come from the same source reports, the Table 2 results are in-domain rather than off-the-shelf. The authors should clarify the exact overlap and rerun the evaluation with a truly held-out benchmark.
  3. [Section 4.3 and Figure 1] The 'average score' used for the 2% gap is an unweighted mean over heterogeneous metric rows (F1, MicroF1, EntityF1, Exact Match, Rouge-1, BERTScore) in Table 1, with no task-level weighting, confidence intervals, or significance tests. The GPT-4 and Gemini numbers are imported from FinBen (footnote *), not re-evaluated under the same prompt template, decoding parameters, or evaluation harness, and the comparison is between fine-tuned SusGen-GPT and zero-shot baselines. These factors make the headline difference not a well-defined quantity; an apples-to-apples evaluation with identical prompts and settings is needed.
  4. [Section 4.3 and Table 2] The TCFD-Bench evaluation includes only SusGen-GPT and ChatReport. No GPT-4 or Gemini results are reported on TCFD-Bench, so the abstract's claim that SusGen-GPT surpasses all models except GPT-4 is not supported for sustainability report generation. The authors should run the same closed-source baselines on TCFD-Bench or qualify the claim.
minor comments (5)
  1. [Figure 1] Figure 1 lacks a description of how the 'average score' is computed; please add the formula and per-task metric definitions to the caption.
  2. [Section 4.4] Section 4.4 contains a typo: 'Instantce' should be 'instance'.
  3. [Table 1 caption] The caption of Table 1 says 'zero-shot and few-shot performance,' but SusGen-GPT is fine-tuned; clarify the inference setting for every model.
  4. [Appendix C.1] Appendix C.1 would benefit from a column explaining the relationship between Train, Test, and Final, especially for rows where Train is marked '✗' but Test and Final are nonzero.
  5. [Limitations] The Limitations section acknowledges the lack of expert evaluation and the restricted TCFD template diversity; these caveats should be more explicitly reflected in the conclusions about report generation quality.

Circularity Check

2 steps flagged · score 8.0 of 10

Table 3 shows FinBen test splits inside the SusGen-30K training set, so the '0.02 below GPT-4' average is a memorization score, not a held-out prediction.

  1. fitted input called prediction [Section 4.3 / Table 1 vs Appendix C.1 / Table 3]
    "flare-fomc (Xie and et. al., 2024) ✗ EN 496 496 valid188 flare-tatqa (Xie and et. al., 2024) ✗ EN 1,668 1,668 flare-edtsum-test (Xie and et. al., 2024) ✗ EN 2000 2000 flare-causal20-sc-test (Xie and et. al., 2024) ✗ EN 8,628 8,628 flare-finer-ord (Xie and et. al., 2024) ✗ EN 1,075 1,075 flare-mlesg-en (Xie and et. al., 2024) ✗ EN 300 300 flare-fiqasa (Xie and et. al., 2024) 750 EN 235 235 valid188"

    These flare-* rows are FinBen benchmark test splits, and FiQASA, FOMC, TATQA, EDTSUM, SC, FINER-ORD and MLESG are exactly the datasets scored in Table 1. The Final column is the amount used to build SusGen-30K, and SusGen-GPT is supervised-fine-tuned on SusGen-30K. For each row, Final equals Test, so the exact test instances are in the training set. The Table 1 scores, and the resulting 'only 0.02 below GPT-4' average in the Abstract and Section 4.3, are therefore forced by construction: they measure reproduction of training examples, not prediction on unseen financial NLP data.

  2. self definitional [Section 4.2, Benchmarks & Baseline Models]
    "Text Summarization and Sustainability Report Generation are considered two non-adapted tasks because our training set does not explicitly include them, while the other six tasks are regarded as adapted tasks."

    This defines 'non-adapted'/'off-the-shelf' by the task's absence from the training set, but the paper's own construction table (Table 3) lists SUM flare-edtsum-test with Final=2,000 and ESG-QA TCFD_QA with Final=1,669 as part of SusGen-30K. EDTSUM is the summarization benchmark evaluated in Table 1, and TCFD_QA is built from the same TCFD-report material used to construct TCFD-Bench. Thus the claim that the two off-the-shelf tasks are absent from training is contradicted by the paper's own data table; the 'non-adapted' evaluation is circular by definition.

full rationale

The central quantitative claim is not a valid generalization measurement. Appendix C.1/Table 3 is explicitly the composition of SusGen-30K, the dataset used to train SusGen-GPT; several rows are FinBen benchmark test splits with Final counts equal to their Test counts (flare-fomc 496/496, flare-tatqa 1,668/1,668, flare-edtsum-test 2,000/2,000, flare-causal20-sc-test 8,628/8,628, flare-finer-ord 1,075/1,075, flare-mlesg-en 300/300, flare-fiqasa 235/235). Those are the same datasets scored in Table 1, which supplies the Abstract's 'average score only 0.02 below GPT-4' and Section 4.3's 'near-GPT-4 performance' conclusion. Because the model was fine-tuned on these exact test instances, the headline numbers are memorization scores, not predictions. Independently, Section 4.2's statement that Text Summarization and Sustainability Report Generation are non-adapted is contradicted by Table 3, which includes EDTSUM and TCFD_QA in the final training mix; the same 14 TCFD reports used to seed TCFD-Bench also feed the training data. This is an internal inconsistency between the paper's own data table and its evaluation claim, not a matter of consensus or self-citation. Score 8 is warranted because the core efficiency result reduces, by the paper's own construction table, to in-sample fit rather than held-out prediction.

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

The paper introduces no new physical or mathematical entities. It relies on standard NLP evaluation metrics, on the accuracy of LLM-generated data in TCFD-Bench, and on the comparability of scores imported from a prior benchmark paper. The most significant unstated assumption is that the adapted tasks' training and test splits are cleanly separated, which the dataset table does not clearly confirm.

assumptions (4)
  • domain assumption Automatic metrics (ROUGE, BERTScore, METEOR, BLEU) adequately capture quality of generated sustainability reports.
    Section 3.4 uses these as the only report-generation evaluation; no human validation is reported, and the authors note this limitation in the Limitations section.
  • domain assumption GPT-4o-generated contexts and Mistral-generated instructions accurately represent TCFD report content without introducing errors.
    Section 3.2 describes using GPT-4o to generate context and Mistral 7B to generate instructions for TCFD-Bench; no verification of correctness is described.
  • domain assumption The GPT-4 and Gemini scores imported from FinBen (Xie et al., 2024) are directly comparable to the authors' own evaluation runs.
    Section 4.3 marks GPT-4 and Gemini results with '*' and states they come from FinBen; the prompts and decoding settings may differ from this paper's setup.
  • domain assumption The FinQA and other adapted benchmarks' training splits used in SusGen-30K are disjoint from the test splits used in evaluation.
    Table 3 lists training quantities from the same benchmark families; the paper never explicitly confirms train/test separation for the adapted tasks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SusGen-GPT: A Data-Centric LLM for Financial NLP and Sustainability Report Generation." pith.science (2026). https://pith.science/paper/6YJ3EMQM

@misc{pith2026241210906,
  author       = {Pith},
  title        = {Pith review of: SusGen-GPT: A Data-Centric LLM for Financial NLP and Sustainability Report Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6YJ3EMQM}},
  note         = {Machine review of arXiv:2412.10906}
}
read the original abstract

The rapid growth of the financial sector and the rising focus on Environmental, Social, and Governance (ESG) considerations highlight the need for advanced NLP tools. However, open-source LLMs proficient in both finance and ESG domains remain scarce. To address this gap, we introduce SusGen-30K, a category-balanced dataset comprising seven financial NLP tasks and ESG report generation, and propose TCFD-Bench, a benchmark for evaluating sustainability report generation. Leveraging this dataset, we developed SusGen-GPT, a suite of models achieving state-of-the-art performance across six adapted and two off-the-shelf tasks, trailing GPT-4 by only 2% despite using 7-8B parameters compared to GPT-4's 1,700B. Based on this, we propose the SusGen system, integrated with Retrieval-Augmented Generation (RAG), to assist in sustainability report generation. This work demonstrates the efficiency of our approach, advancing research in finance and ESG.

Figures

Figures reproduced from arXiv: 2412.10906 by the authors.

Figure 1
Figure 1. An overview of model comparison with both [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of the SusGen System Pipeline. Context Text of the company’s introduction and given topic. Instruction Initially created manually and diversified by GPT. Question Question from the disclosure report on the TCFD topic. Output Answer extracted from the TCFD report. Company Annual Report GPT-4o Manual TCFD Report Context: Wolfspeed, Inc. is a leading global company specializing in the development and productio… view at source ↗
Figure 4
Figure 4. The pipeline of SusGen-30K data construc [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figures from the paper (5 more)
Figure 3
Figure 3. Figure 3: The data construction pipeline of TCFD￾Bench, illustrated with an example extracted and pro￾cessed from the Wolfspeed_2022.pdf reports. 3.3 Statistics In this section, we present the statistical informa￾tion about our training dataset, SusGen-30K. As illustrated in [P…
Figure 5
Figure 5. Figure 5: SusGen-30K Category Distribution. Highlight [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Ablation study results of our models SusGen-GPT trained on 10k, 20k, and 30k subsets of our dataset [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: A generated Sustainability Report Example. A comparison of the output obtained by our SusGen-GPT [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Token length distribution across the 9 sections of S [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 12 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

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

  3. [3]

    Aguda, Suchetha Siddagangappa, Elena Kochkina, Simerjot Kaur, Dongsheng Wang, and Charese Smiley

    Toyin D. Aguda, Suchetha Siddagangappa, Elena Kochkina, Simerjot Kaur, Dongsheng Wang, and Charese Smiley. 2024. https://aclanthology.org/2024.lrec-main.885 Large language models as financial data annotators: A study on effectiveness and efficiency . In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources...

    2024

  4. [4]

    Ebtesam Almazrouei, Hamza Alobeidli, Abdulaziz Alshamsi, Alessandro Cappelli, Ruxandra Cojocaru, M \'e rouane Debbah, \'E tienne Goffinet, Daniel Hesslow, Julien Launay, Quentin Malartic, et al. 2023. The falcon series of open language models. arXiv preprint arXiv:2311.16867

    arXiv 2023

  5. [5]

    Julio Cesar Salinas Alvarado, Karin Verspoor, and Timothy Baldwin. 2015. Domain adaption of named entity recognition to support credit risk assessment. In Proceedings of the Australasian Language Technology Association Workshop 2015, pages 84--90

    work page 2015

  6. [6]

    Satanjeev Banerjee and Alon Lavie. 2005. Meteor: An automatic metric for mt evaluation with improved correlation with human judgments. In Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summarization, pages 65--72

    2005

  7. [7]

    Dan Biderman, Jose Gonzalez Ortiz, Jacob Portes, Mansheej Paul, Philip Greengard, Connor Jennings, Daniel King, Sam Havens, Vitaliy Chiley, Jonathan Frankle, et al. 2024. Lora learns less and forgets less. arXiv preprint arXiv:2405.09673

    arXiv 2024

  8. [8]

    Marco Bronzini, Carlo Nicolini, Bruno Lepri, Andrea Passerini, and Jacopo Staiano. 2024. Glitter or gold? deriving structured insights from sustainability reports via large language models. EPJ Data Science, 13(1):41

    work page 2024

Show all 52 references
  1. [9]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877--1901

    2020

  2. [10]

    Chung-Chi Chen, Yu-Min Tseng, Juyeon Kang, Ana \" s Lhuissier, Min-Yuh Day, Teng-Tsai Tu, and Hsin-Hsi Chen. 2023. Multi-lingual esg issue identification. In Proceedings of the Fifth Workshop on Financial Technology and Natural Language Processing and the Second Multimodal AI ...

    2023

  3. [11]

    Zhiyu Chen, Wenhu Chen, Charese Smiley, Sameena Shah, Iana Borova, Dylan Langdon, Reema Moussa, Matt Beane, Ting-Hao Huang, Bryan R Routledge, et al. 2021. Finqa: A dataset of numerical reasoning over financial data. In Proceedings of the 2021 Conference on Empirical Methods i...

    2021

  4. [12]

    Zhiyu Chen, Shiyang Li, Charese Smiley, Zhiqiang Ma, Sameena Shah, and William Yang Wang. 2022. https://arxiv.org/abs/2210.03849 Convfinqa: Exploring the chain of numerical reasoning in conversational finance question answering . Preprint, arXiv:2210.03849

    2022 arXiv

  5. [13]

    Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168

    2021 arXiv

  6. [14]

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2024. Qlora: Efficient finetuning of quantized llms. Advances in Neural Information Processing Systems, 36

    2024

  7. [15]

    Abhimanyu Dubey and Abhinav Jauhri et. al. 2024. https://arxiv.org/abs/2407.21783 The llama 3 herd of models . Preprint, arXiv:2407.21783

    2024 arXiv

  8. [16]

    Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, Lélio Renard Lavaud, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas ...

    2023 arXiv

  9. [17]

    Rasmus J rgensen, Oliver Brandt, Mareike Hartmann, Xiang Dai, Christian Igel, and Desmond Elliott. 2023. Multifin: A dataset for multilingual financial nlp. In Findings of the Association for Computational Linguistics: EACL 2023, pages 864--879

    2023

  10. [18]

    Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B. Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. 2020. https://arxiv.org/abs/2001.08361 Scaling laws for neural language models . Preprint, arXiv:2001.08361

    2020 arXiv

  11. [19]

    Yang Lei, Jiangtong Li, Dawei Cheng, Zhijun Ding, and Changjun Jiang. 2024. https://arxiv.org/abs/2311.05812 Cfbenchmark: Chinese financial assistant benchmark for large language model . Preprint, arXiv:2311.05812

    2024 arXiv

  12. [20]

    Chen Li, Weiqi Wang, Jingcheng Hu, Yixuan Wei, Nanning Zheng, Han Hu, Zheng Zhang, and Houwen Peng. 2024. https://arxiv.org/abs/2403.04706 Common 7b language models already possess strong math capabilities . Preprint, arXiv:2403.04706

    2024 arXiv

  13. [21]

    Chin-Yew Lin. 2004. Rouge: A package for automatic evaluation of summaries. In Text summarization branches out, pages 74--81

    2004

  14. [22]

    Xiao-Yang Liu, Guoxuan Wang, Hongyang Yang, and Daochen Zha. 2023. Data-centric fingpt: Democratizing internet-scale data for financial large language models. NeurIPS Workshop on Instruction Tuning and Instruction Following

    2023

  15. [23]

    Ilya Loshchilov and Frank Hutter. 2019. https://arxiv.org/abs/1711.05101 Decoupled weight decay regularization . Preprint, arXiv:1711.05101

    2019 arXiv

  16. [24]

    Alexandra Luccioni, Emily Baylor, and Nicolas Duchene. 2020. https://arxiv.org/abs/2011.08073 Analyzing sustainability reports using natural language processing . Preprint, arXiv:2011.08073

    2020 arXiv

  17. [25]

    Macedo Maia, Siegfried Handschuh, Andr \'e Freitas, Brian Davis, Ross McDermott, Manel Zarrouk, and Alexandra Balahur. 2018. Www'18 open challenge: financial opinion mining and question answering. In Companion proceedings of the the web conference 2018, pages 1941--1942

    2018

  18. [26]

    Dominique Mariko, Hanna Abi Akl, Estelle Labidurie, Stephane Durfort, Hugues De Mazancourt, and Mahmoud El-Haj. 2020. Financial document causality detection shared task (fincausal 2020). arXiv preprint arXiv:2012.02505

    2020 arXiv

  19. [27]

    Corentin Masson and Patrick Paroubek. 2024. https://aclanthology.org/2024.lrec-main.578 Evaluating topic model on asymmetric and multi-domain financial corpus . In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evalu...

    2024

  20. [28]

    Jingwei Ni, Julia Bingler, Chiara Colesanti-Senni, Mathias Kraus, Glen Gostlow, Tobias Schimanski, Dominik Stammbach, Saeid Ashraf Vaghefi, Qian Wang, Nicolas Webersinke, Tobias Wekhof, Tingyu Yu, and Markus Leippold. 2023. https://doi.org/10.18653/v1/2023.emnlp-demo.3 CHATREP...

    2023 doi

  21. [29]

    OpenAI. 2023 a . https://cdn.openai.com/papers/gpt-4.pdf Gpt-4 technical report . Technical report, OpenAI

    2023

  22. [30]

    OpenAI. 2023 b . https://arxiv.org/abs/2303.08774 Gpt-4 technical report . Preprint, arXiv:2303.08774

    2023 arXiv

  23. [31]

    OpenAI. 2023 c . Instructgpt: Ai for generating instructions. https://openai.com/research/instructgpt/

    2023

  24. [32]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 3...

    2022

  25. [33]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting of the Association for Computational Linguistics, pages 311--318

    2002

  26. [34]

    Agam Shah, Suvan Paturi, and Sudheer Chava. 2023 a . https://doi.org/10.18653/v1/2023.acl-long.368 Trillion dollar words: A new financial dataset, task & market analysis . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Lon...

    2023 doi

  27. [35]

    Agam Shah, Ruchit Vithani, Abhinav Gullapalli, and Sudheer Chava. 2023 b . Finer: Financial named entity recognition dataset and weak-supervision model. arXiv preprint arXiv:2302.11157

    2023 arXiv

  28. [36]

    Raj Sanjay Shah, Kunal Chawla, Dheeraj Eidnani, Agam Shah, Wendi Du, Sudheer Chava, Natraj Raman, Charese Smiley, Jiaao Chen, and Diyi Yang. 2022. https://arxiv.org/abs/2211.00083 When flue meets flang: Benchmarks and large pre-trained language model for financial domain . Pre...

    2022 arXiv

  29. [37]

    Soumya Sharma, Tapas Nayak, Arusarka Bose, Ajay Kumar Meena, Koustuv Dasgupta, Niloy Ganguly, and Pawan Goyal. 2022. Finred: A dataset for relation extraction in financial domain. In Companion Proceedings of the Web Conference 2022, pages 595--597

    2022

  30. [38]

    Hashimoto

    Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B. Hashimoto. 2023. Stanford alpaca: An instruction-following llama model. https://github.com/tatsu-lab/stanford_alpaca

    2023

  31. [39]

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805

    2023 arXiv

  32. [40]

    Hugo Touvron and Lavril et. al. 2023 a . Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971

    2023 arXiv

  33. [41]

    Hugo Touvron and Louis Martin et. al. 2023 b . https://arxiv.org/abs/2307.09288 Llama 2: Open foundation and fine-tuned chat models . Preprint, arXiv:2307.09288

    2023 arXiv

  34. [42]

    Neng Wang, Hongyang Yang, and Christina Dan Wang. 2023 a . Fingpt: Instruction tuning benchmark for open-source large language models in financial datasets. NeurIPS Workshop on Instruction Tuning and Instruction Following

    2023

  35. [43]

    Neng Wang, Hongyang Yang, and Christina Dan Wang. 2023 b . https://arxiv.org/abs/2310.04793 Fingpt: Instruction tuning benchmark for open-source large language models in financial datasets . Preprint, arXiv:2310.04793

    2023 arXiv

  36. [44]

    Shijie Wu, Ozan Irsoy, Steven Lu, Vadim Dabravolski, Mark Dredze, Sebastian Gehrmann, Prabhanjan Kambadur, David Rosenberg, and Gideon Mann. 2023. https://arxiv.org/abs/2303.17564 Bloomberggpt: A large language model for finance . Preprint, arXiv:2303.17564

    2023 arXiv

  37. [45]

    Qianqian Xie and Weiguang Han et. al. 2023. https://arxiv.org/abs/2306.05443 Pixiu: A large language model, instruction data and evaluation benchmark for finance . Preprint, arXiv:2306.05443

    2023 arXiv

  38. [46]

    Qianqian Xie and Weiguang Han et. al. 2024. https://arxiv.org/abs/2402.12659 The finben: An holistic financial benchmark for large language models . Preprint, arXiv:2402.12659

    2024 arXiv

  39. [47]

    Hongyang Yang, Xiao-Yang Liu, and Christina Dan Wang. 2023. Fingpt: Open-source financial large language models. FinLLM Symposium at IJCAI 2023

    2023

  40. [48]

    Wei Jie Yeo, Teddy Ferdinan, Przemyslaw Kazienko, Ranjan Satapathy, and Erik Cambria. 2024. Self-training large language models through knowledge detection. arXiv preprint arXiv:2406.11275

    2024 arXiv

  41. [49]

    Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q Weinberger, and Yoav Artzi. 2019. Bertscore: Evaluating text generation with bert. arXiv preprint arXiv:1904.09675

    2019 arXiv

  42. [50]

    Zhihan Zhou, Liqian Ma, and Han Liu. 2021. https://arxiv.org/abs/2105.12825 Trade the event: Corporate events detection for news-based event-driven trading . Preprint, arXiv:2105.12825

    2021 arXiv

  43. [51]

    Fengbin Zhu, Wenqiang Lei, Youcheng Huang, Chao Wang, Shuo Zhang, Jiancheng Lv, Fuli Feng, and Tat-Seng Chua. 2021. Tat-qa: A question answering benchmark on a hybrid of tabular and textual content in finance. arXiv preprint arXiv:2105.07624

    2021 arXiv

  44. [52]

    Yi Zou, Mengying Shi, Zhongjie Chen, Zhu Deng, ZongXiong Lei, Zihan Zeng, Shiming Yang, HongXiang Tong, Lei Xiao, and Wenwen Zhou. 2023. Esgreveal: An llm-based approach for extracting structured data from esg reports. arXiv preprint arXiv:2312.17264

    2023 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.