Pith. sign in

REVIEW 4 major objections 5 minor 26 references

LLM-based Affective Text Generation Quality Based on Different Quantization Values

T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read Quantizing LLMs from 32 to 8 bits cuts GPU memory by about 76% for affective text generation, with larger quantized models often beating smaller full-precision ones at similar memory.

desk verdict A useful memory-quantization measurement whose F1-based trade-off is compromised by an impossible F1 value and an inference-time claim contradicted by its own table; deserving of a major-revision review. read the letter →

arxiv 2501.19317 v1 pith:KCFQXZOL submitted 2025-01-31 cs.CL

Yarik Menchaca Resendiz , Roman Klinger This is my paper · ORCID
classification cs.CL
keywords quantizationaffectivetextgenerationemotionclassificationGPUmemoryinferencetimelargelanguagemodelsLlama-2Mistral
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 what happens to affective text generation when large language models are quantized to 8, 16, or 32 bits. It finds that dropping from 32 to 8 bits cuts GPU memory usage by about 76% across five open-weight models, at the price of roughly doubling inference time. The emotion-classification F1 of the generated text falls by up to 10 percentage points for the largest model (Llama-2-70B) but can rise by up to 10 points for smaller models (Llama-2-7B). A key result is that larger models at 8 bits generally match or beat smaller models at 16 or 32 bits on both the emotion classifier and on LLM-based text-quality scores, while using about the same memory. This points toward a practical recipe for running capable affective text generation on memory-limited hardware: choose the largest model you can fit at 8 bits rather than a smaller model at full precision.

What carries the argument

The central object is the precision-bit setting (8, 16, or 32) applied to five open-weight LLMs, where quantization means reducing the number of bits used to store each model weight. The evaluation machinery is an emotion classifier (RoBERTa fine-tuned on the ISEAR dataset) that scores whether generated text expresses the target emotion, plus GPU RAM and inference-time measurements, plus GPT-4o and GPT-3.5 Likert ratings for text quality dimensions like coherence, fluency, and plausibility.

What would settle it

Have human annotators label whether each generated sentence expresses the target emotion, and compare the resulting emotion accuracy to the classifier's F1 across the 15 configurations; if the human-based ranking does not match the classifier's, the reported trade-off is at least partly an artifact of the classifier.

Watch

Extended reading notes

Core claim

The paper reports that lowering precision from 32 to 8 bits reduces GPU memory by roughly 76% across all tested configurations, while changing the F1 of an emotion classifier on generated text by up to −10 percentage points for the largest model (Llama-2-70B) and +10 points for smaller models (Llama-2-7B). At the same time, the larger models at 8 bits matched or beat smaller models at 16 or 32 bits in both the classifier's F1 and in GPT-4o text-quality ratings, despite using similar memory. That means an 8-bit Llama-2-70B can be a practical substitute for a 16- or 32-bit 7B or 13B model on memory-limited hardware, with the caveat that inference takes roughly twice as long.

Load-bearing premise

The automatic emotion classifier used to score whether generated text expresses the target emotion is treated as ground truth, and if it is biased toward certain model sizes or precisions, the reported F1 trade-offs could be artifacts.

Editorial extensions

If this is right

  • An 8-bit Llama-2-70B can deliver better emotion-text quality than a 16- or 32-bit 7B or 13B model while using roughly the same GPU memory, making it a practical choice on memory-limited GPUs.
  • Quantizing from 32 to 8 bits consistently saves about 76% of GPU RAM, but most configurations see inference time roughly double.
  • The effect of quantization on emotion correctness is not uniform: some small models improve in F1 after quantization, while large models can lose up to 10 percentage points.
  • Because the evaluation rests on a single emotion classifier and ten seed prompts, the reported F1 numbers are tied to that classifier's notion of 'expressing an emotion' and could shift with a different classifier or prompt set.

Reading between the lines

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

  • If the emotion classifier systematically favors longer or more fluent outputs, the 'larger model at 8 bits beats smaller at full precision' result could be driven by text length or fluency rather than by emotion accuracy; a human study that controls for length would settle this.
  • The same memory-budget logic likely generalizes to other generation tasks such as summarization or dialogue, but the specific ±10 pp F1 range and the 76% memory figure are tied to this model set, this classifier, and these ten prompts.
  • A cheap extension would be to test whether the 'bigger and quantized beats smaller and precise' ordering holds when the emotion classifier is replaced by human evaluation on the same generated sentences; the paper's Limitation section effectively invites this check.
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 studies how post-training quantization (8, 16, and 32 bits) affects affective text generation in five open-weight LLMs (Llama-2 7B/13B/70B, Mistral-7B, Mixtral-8x7B). Ten seed prompts are combined with seven emotion categories to generate 210 sentences per model-precision configuration, which are scored by a RoBERTa emotion classifier fine-tuned on ISEAR for whether they express the target emotion (F1), and by GPT-4o and GPT-3.5 judges on five text-quality dimensions. The central claim is that quantizing to 8 bits reduces GPU memory by about 76%, costs up to 10 percentage points of F1 for larger models, gains up to 10 points for smaller models, roughly doubles inference time, and yields text quality in which larger models at lower quantization generally outperform smaller, higher-precision models at similar memory.

Significance. If the findings are correct, the paper offers a practical, actionable trade-off map for deploying LLM-based affective generation on memory-constrained hardware, a question that is relevant and underexplored. The paper has concrete strengths: it systematically covers 15 model-precision configurations, reports hardware measurements (GPU RAM, inference time) that are directly useful, provides a clear and reproducible-looking protocol, and makes code available in principle. The main quantitative claims, however, rest on an emotion classifier whose reported evaluation contains an internal inconsistency, and the F1 comparisons lack any measures of uncertainty. The text-quality conclusions also rely on LLM judges without human validation and are partly contradicted by the paper's own tables. The topic is timely, but the evidence as presented does not yet support the headline trade-off claims.

major comments (4)
  1. [Table 1, §2.4] Table 1 reports for the Joy class a precision of 0.91 and a recall of 0.92, but an F1 score of 0.98. The harmonic mean is 2*0.91*0.92/(0.91+0.92) = 0.915, which is 0.065 away from the reported value and cannot be explained by rounding. Because this classifier computes every F1 score in Table 2, the central quantitative claim about up to ±10 pp changes in F1 inherits this unreliability. The authors must correct the classifier evaluation (or the reporting error) and recompute all downstream F1 values before the trade-off claim can be assessed.
  2. [Table 2, §3.1] The F1 differences that drive the main conclusion — e.g., Llama-2-7b 32-bit at .63 vs 8-bit at .73, or Llama-2-70b 32-bit at .74 vs 8-bit at .64 — are reported as point estimates only. Each cell is based on 210 generated sentences across 7 emotion classes and 10 seed prompts, yet no variance, confidence interval, or significance test is given. With this sample size, differences on the order of 0.10 could easily arise from sampling noise. The paper should provide per-emotion F1 breakdowns, bootstrap intervals, or a significance test (e.g., paired across prompts) to support the claimed 10 pp increases and decreases.
  3. [§3.4, Table 2] The claim that 'generating an instance with 8-bit quantization takes about twice as long as with 32-bit quantization' is not supported by the paper's own data. The 8-bit to 32-bit inference time ratios in Table 2 are 1.32 for Llama-2-7b, 1.39 for Llama-2-13b, 0.67 for Llama-2-70b, 2.11 for Mistral-7B, and 2.33 for Mixtral-8x7B. Similarly, the statement in §3.4 that 'as quantization increases, inference time also increases' is contradicted by Llama-2-70b, whose 8-bit runtime (34.93 s) is lower than its 32-bit runtime (51.93 s). This trade-off component needs to be re-analyzed and the claims revised or qualified.
  4. [§3.2, Table 3] The conclusion that 'larger models at lower quantization levels generally outperform smaller, higher-precision models -- while requiring similar memory' is not uniformly supported by Table 3. Mistral-7B at any quantization level scores higher than Llama-2-70B on all five reported quality dimensions (e.g., fluency 4.6 vs 3.6 at 8 bits), so the 'generally' claim requires a caveat or a more careful ranking analysis. In addition, the emotion-correctness evaluation is entirely delegated to an automatic classifier with macro-F1 0.75 and a Shame F1 of only 0.52; no human evaluation is reported to confirm that the generated text actually expresses the requested emotion. The F1-based trade-off could be an artifact of classifier bias correlated with model size or precision. The authors should either add a human validation sample or explicitly limit the claim to classifier-measured emotion fidelity.
minor comments (5)
  1. [Appendix A] The text refers to 'Table ??' when discussing the GPT-3.5 evaluation; this placeholder should be replaced with a reference to Table 4.
  2. [§2.5, Tables 3 and 4] Section 2.5 lists six text-quality dimensions, including Human Likeness, but Tables 3 and 4 report only five columns (Fluency, Native Speaker Likeness, Coherency, Plausibility, Grammar) and omit Human Likeness. Please align the text with the tables or add the missing dimension.
  3. [§2.3] The sentence 'The inference tasks for the LLMs models are performed and evenly distributed across a GPU cluster equipped with 8 NVIDIA RTX A6000 GPUs, for 10 times per model configuration' is ambiguous: it is unclear whether the entire generation is repeated 10 times per model configuration and, if so, how the 210-sentence counts and the reported F1 values aggregate these repetitions.
  4. [§3.1] The claim that quantization reduces performance 'by an average of 3.6 pp across all configurations' is not derivable from Table 2. For example, the mean F1 across the five models is 0.71 at 8 bits, 0.70 at 16 bits, and 0.69 at 32 bits, which gives an average drop of roughly 1.6 pp from 8 to 32 bits and less from 16 to 32 bits. Please specify the baseline and computation for the 3.6 pp figure.
  5. [Table 1] The macro-averaged precision, recall, and F1 are all reported as .75, which is consistent with rounding, but the per-class F1 values should be recomputed after correcting the Joy F1 and the table regenerated with consistent two-decimal rounding.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation; the paper is an empirical benchmark with independent classifier scores and direct hardware measurements, and the authors' prior work appears only as background.

full rationale

The paper's central claims are empirical rather than derived: GPU memory and inference time are direct measurements, and the affective F1 scores come from a separately fine-tuned RoBERTa emotion classifier applied to generated text. The text-quality ratings come from GPT-4o and GPT-3.5 prompts, not from the quantization pipeline. The only author self-citations in the load-bearing path are background references: Menchaca Resendiz and Klinger (2023) is cited in the introduction as prior work on emotion-conditioned generation, and Troiano et al. (2023) is cited only to say the classifier's scores are 'comparable to previous experiments'; neither supplies a fitted parameter, a uniqueness theorem, or an ansatz that determines the reported trade-offs. No equation in the paper defines an output quantity in terms of the quantity it purports to predict. The apparent arithmetic inconsistency in Table 1 (Joy precision 0.91, recall 0.92, F1 0.98) is a correctness or reporting concern, not a circularity concern, because it does not show that the Table 2 F1 values or the memory-savings conclusion were constructed from the inputs. Overall, the derivation chain is self-contained against external benchmarks.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No model parameters are fitted in this paper; the numbers the conclusions depend on are evaluation choices and decoding hyperparameters chosen by the authors without sensitivity analysis or release. The main load-bearing assumptions are that the ISEAR-trained classifier is a valid judge of emotion correctness and that LLM quality ratings approximate human judgments.

free parameters (5)
  • decoding temperature = 0.7
    Set by hand in Section 2.3; no sensitivity analysis, but it shapes all generated texts and therefore the F1 and quality scores.
  • nucleus sampling top-p = 0.7
    Set by hand in Section 2.3; interacts with the claimed beam search in a way that is not described.
  • beam size = 30
    Set by hand in Section 2.3; changes output diversity and relevance.
  • emotion classifier epochs = 10
    Training choice in Section 2.4; determines the classifier that produces all F1 scores.
  • emotion classifier batch size = 5
    Training choice in Section 2.4; contributes to the classifier's reported quality.
assumptions (4)
  • domain assumption The ISEAR-fine-tuned RoBERTa classifier, with macro-F1 0.75, is a valid automatic judge of whether generated text expresses the target emotion.
    Used in Section 2.4 and for all F1 scores in Table 2; no human evaluation of generated outputs is reported.
  • domain assumption The 10 seed prompts and 7 ISEAR emotion categories are representative of affective text generation.
    Defined in Section 2.1; the conclusions may not generalize to other prompts, languages, or emotion taxonomies.
  • domain assumption GPT-4o and GPT-3.5 Likert ratings approximate human text-quality judgments.
    Invoked in Section 2.5; no human alignment check on these specific outputs is provided.
  • domain assumption Standard post-training quantization to 8/16/32 bits preserves the model's generation behavior sufficiently for comparison.
    Section 2.2 gives precision bits but no quantization method, library, or parameters, so the comparison assumes a conventional implementation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLM-based Affective Text Generation Quality Based on Different Quantization Values." pith.science (2026). https://pith.science/paper/KCFQXZOL

@misc{pith2026250119317,
  author       = {Pith},
  title        = {Pith review of: LLM-based Affective Text Generation Quality Based on Different Quantization Values},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KCFQXZOL}},
  note         = {Machine review of arXiv:2501.19317}
}
read the original abstract

Large language models exhibit a remarkable capacity in language generation and comprehension. These advances enable AI systems to produce more human-like and emotionally engaging text. However, these models rely on a large number of parameters, requiring significant computational resources for training and inference. In some scenarios, accessing these resources can be challenging (e.g., budget or hardware limitations). Techniques like reducing precision bits can make models more memory-efficient, reducing the computational resources needed, at the cost of reduced accuracy. This paper addresses the trade-off between different quantization values, GPU RAM utilization, and text quality in affective text generation (e.g., "I really enjoy running in the snow-covered forest"). To evaluate, we use an emotion classifier and ten seed prompts to generate affective text. We test three setups of precision bits (8, 16, and 32) across five open-weight language models from two different families. Our findings demonstrate that bit reductions lead to memory savings, achieving a reduction of 76%. However, this optimization comes with a trade-off, leading to a decrease of up to 10 pp in F1 score for larger models and an increase of 10 pp for smaller models, along with roughly double the inference time. In terms of text quality, larger models at lower quantization levels generally outperform smaller, higher-precision models -- while requiring similar memory.

Figures

Figures reproduced from arXiv: 2501.19317 by the authors.

Figure 1
Figure 1. F1-score vs. GPU memory usage for different models and configurations. with higher precision – for example, Llama-2-70b at 8 bits outperforms Llama-2-7b or Llama-2-13b at 16 or 32 bits – while requiring similar memory. 3.3 GPU Usage vs. F1-score [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. GPU usage (GB) vs. inference time (seconds) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 15 canonical work pages

  1. [1]

    URL: " 'urlintro :=

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

  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]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gr...

    work page 2020

  4. [4]

    Cheng-Han Chiang and Hung-yi Lee. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.599 A closer look into using large language models for automatic evaluation . In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 8928--8942, Singapore. Association for Computational Linguistics

    doi:10.18653/v1/2023.findings-emnlp.599 2023

  5. [5]

    Pierre Colombo, Wojciech Witon, Ashutosh Modi, James Kennedy, and Mubbasir Kapadia. 2019. https://doi.org/10.18653/v1/N19-1374 Affect-driven dialog generation . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) , pages 3734--3...

    work page doi:10.18653/v1/n19-1374 2019

  6. [6]

    Matthieu Courbariaux, Yoshua Bengio, and Jean-Pierre David. 2015. https://api.semanticscholar.org/CorpusID:1518846 Binaryconnect: Training deep neural networks with binary weights during propagations . In Neural Information Processing Systems

    work page 2015

  7. [7]

    Fuwei Cui, Hui Di, Lei Shen, Kazushige Ouchi, Ze Liu, and Jinan Xu. 2022. https://link.springer.com/article/10.1007/s10489-021-02683-x Modeling semantic and emotional relationship in multi-turn emotional conversations using multi-task learning . Applied Intelligence, 52(4):4663--4673

    work page doi:10.1007/s10489-021-02683-x 2022

  8. [8]

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2023. https://proceedings.neurips.cc/paper_files/paper/2023/file/1feb87871436031bdc0f2beaa62a049b-Paper-Conference.pdf Qlora: Efficient finetuning of quantized llms . In Advances in Neural Information Processing Systems, volume 36, pages 10088--10115. Curran Associates, Inc

    2023

Show all 26 references
  1. [9]

    Sayan Ghosh, Mathieu Chollet, Eugene Laksana, Louis-Philippe Morency, and Stefan Scherer. 2017. https://doi.org/10.18653/v1/P17-1059 Affect- LM : A neural language model for customizable affective text generation . In Proceedings of the 55th Annual Meeting of the Association f...

    2017 doi

  2. [10]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. https://openreview.net/forum?id=nZeVKeeFYf9 Lo RA : Low-rank adaptation of large language models . In International Conference on Learning Representations

    2022

  3. [11]

    Benoit Jacob, Skirmantas Kligys, Bo Chen, Menglong Zhu, Matthew Tang, Andrew Howard, Hartwig Adam, and Dmitry Kalenichenko. 2018. https://doi.org/10.1109/CVPR.2018.00286 Quantization and training of neural networks for efficient integer-arithmetic-only inference . In 2018 IEEE...

    2018

  4. [12]

    Albert Q Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al. 2023. Mistral 7b. arXiv preprint arXiv:2310.06825

    2023 arXiv

  5. [13]

    Albert Q Jiang, Alexandre Sablayrolles, Antoine Roux, Arthur Mensch, Blanche Savary, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Emma Bou Hanna, Florian Bressand, et al. 2024. Mixtral of experts. arXiv preprint arXiv:2401.04088

    2024 arXiv

  6. [14]

    Yang Liu, Dan Iter, Yichong Xu, Shuohang Wang, Ruochen Xu, and Chenguang Zhu. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.153 G -eval: NLG evaluation using gpt-4 with better human alignment . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language ...

    2023 doi

  7. [15]

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. https://arxiv.org/abs/1907.11692 Roberta: A robustly optimized bert pretraining approach . arXiv preprint arXiv:1907.11692

    2019 arXiv

  8. [16]

    Umar Maqsud. 2015. https://doi.org/10.18653/v1/W15-2922 Synthetic text generation for sentiment analysis . In Proceedings of the 6th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis, pages 156--161, Lisboa, Portugal. Association for Com...

    2015 doi

  9. [17]

    Yarik Menchaca Resendiz and Roman Klinger. 2023. https://doi.org/10.18653/v1/2023.inlg-main.26 Affective natural language generation of event descriptions through fine-grained appraisal conditions . In Proceedings of the 16th International Natural Language Generation Conferenc...

    2023 doi

  10. [18]

    Tong Niu and Mohit Bansal. 2018. https://transacl.org/ojs/index.php/tacl/article/view/1424 Polite dialogue generation without parallel data . Transactions of the Association for Computational Linguistics, 6:373--389

    2018

  11. [19]

    Alec Radford, Karthik Narasimhan, Tim Salimans, and Ilya Sutskever. 2018. https://www.gwern.net/docs/www/s3-us-west-2.amazonaws.com /d73fdc5ffa8627bce44dcda2fc012da638ffb158.pdf Improving language understanding by generative pre-training . OpenAI blog

    2018

  12. [20]

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. 2019. https://d4mucfpksywv.cloudfront.net/better-language-models/language-models.pdf Language models are unsupervised multitask learners . OpenAI blog, 1(8):9

    2019

  13. [21]

    Scherer and Harald G

    Klaus R. Scherer and Harald G. Wallbott. 1994. https://doi.org/10.1037/0022-3514.66.2.310 Evidence for universality and cultural variation of differential emotion response patterning. Journal of personality and social psychology, 66(2):310

    1994 doi

  14. [22]

    Zhenqiao Song, Xiaoqing Zheng, Lu Liu, Mu Xu, and Xuanjing Huang. 2019. https://doi.org/10.18653/v1/P19-1359 Generating responses with a specific emotion in dialog . In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3685--3695, F...

    2019 doi

  15. [23]

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv e-prints, pages arXiv--2307

    2023

  16. [24]

    Enrica Troiano, Laura Oberl\"ander, and Roman Klinger. 2023. https://doi.org/10.1162/coli_a_00461 Dimensional modeling of emotions in text with appraisal theories: Corpus creation, annotation reliability, and prediction . Computational Linguistics, 49(1)

    2023 doi

  17. [25]

    Rui Zhang, Zhenyu Wang, Kai Yin, and Zhenhua Huang. 2019. https://doi.org/10.1109/ACCESS.2019.2931036 Emotional text generation based on cross-domain sentiment transfer . IEEE Access, 7:100081--100089

    2019

  18. [26]

    Hao Zhou, Minlie Huang, Tianyang Zhang, Xiaoyan Zhu, and Bing Liu. 2018. https://doi.org/10.5555/3504035.3504125 Emotional chatting machine: Emotional conversation generation with internal and external memory . In Proceedings of the Thirty-Second AAAI Conference on Artificial ...

    2018

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.