Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Can linguists better understand DNA?

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 Fine-tuning on English sentence-pair data gives 78–89% accuracy on DNA-pair classification, the paper claims.

desk verdict A novel pilot experiment undermined by missing control baselines and best-seed selection; the transfer claim is not yet established. read the letter →

arxiv 2412.07678 v3 pith:5YB5SSCS submitted 2024-12-10 cs.CL q-bio.GN

Wang Liang This is my paper · ORCID
classification cs.CLq-bio.GN
keywords languagecapabilitytransferDNAsequenceclassificationcross-domainmultilingualmodelsPAWS-Xsentence-pairsimilarityGPT-2BERT
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 sets out to show that capabilities a language model learns from natural language can be reused for gene sequences, specifically for judging whether two DNA sequences are similar. After fine-tuning on English sentence-pair similarity data, GPT-2-small reaches 78% accuracy and the multilingual BERT variant reaches 89% on DNA-pair classification with 50-base-pair sequences. The same fine-tuned models perform near chance on DNA-protein pair classification, which the paper reads as evidence that transfer is tied to structural task similarity rather than a general biological understanding. If the claim holds, English-only linguistic supervision could bootstrap biological sequence analysis tasks that currently require labeled genomic data.

What carries the argument

The carrying mechanism is the sentence-pair similarity classification setup borrowed from PAWS-X, a benchmark in which a model must decide whether two sentences are paraphrases. The paper re-uses that exact input-output shape for DNA: two sequences concatenated as one token stream, with a binary label, and a softmax classification head on GPT-2 or BERT. To make the token count comparable, DNA sequences are truncated to 50 base pairs, bringing the input to roughly the same length as English PAWS-X pairs. The paper also identifies shared tokenization (BPE or WordPiece) and structural task similarity as the enabling conditions, and supports this with a mean-nearest-neighbor distance between DNA and English word vectors that drops from 50.09 after pre-training to 29.79 after PAWS-X fine-tuning.

What would settle it

Fine-tune the same models on PAWS-X labels that have been randomly permuted, or run a simple k-mer overlap classifier (for example Jaccard similarity of 3-mers) on the DNA-pair test set; if either matches or exceeds the reported 78–89% accuracy, the result does not demonstrate language capability transfer.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that natural-language sentence-pair training transfers to DNA sequence similarity judgment. The authors construct a DNA-pair classification dataset in which positive pairs are homologous sequences found by BLAST and negative pairs are random sequences of similar length from other organisms, mirroring the PAWS-X paraphrase-pair format. They fine-tune GPT-2 and BERT models on English PAWS-X and then evaluate on DNA pairs, reporting 78% accuracy for GPT-2-small, 84% for English BERT, and 89% for multilingual BERT. Transfer to DNA-protein pair classification fails, with accuracy near 0.5, and the paper attributes the failure to a mismatch between text-paraphrase similarity and genetic coding rules. Additional experiments report that mixed pre-training on English, DNA and protein data improves transfer, and that seed choice affects DNA results much more than natural-language results.

Load-bearing premise

The load-bearing premise is that high accuracy on the DNA-pair test measures transfer of a linguistic capability; since positives are homologous BLAST hits and negatives are random foreign sequences, a model could achieve the reported scores with token-overlap statistics that have nothing to do with language.

Editorial extensions

If this is right

  • English-only fine-tuning on sentence-pair data can yield DNA similarity classifiers with reported accuracy in the high 70s to high 80s, without any DNA-labeled training.
  • The transfer is task-specific: paraphrase-style similarity training does not transfer to DNA-protein coding judgment, so biological tasks must be matched structurally to a natural-language analogue.
  • Pre-training on a mixture of natural language and biological sequences strengthens the transfer effect.
  • Because seed variation can swing DNA accuracy from about 0.52 to 0.79, selecting the best of several fine-tuning runs is important for reliable results.
  • Larger model scale helps only modestly; the smallest GPT-2 is a competitive baseline, implying that scale is not the main driver of transfer.

Reading between the lines

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

  • The DNA-pair task may be solvable by a non-linguistic token-overlap heuristic, since positive pairs are BLAST homologs and negatives are random foreign sequences; without a control baseline, the high accuracy does not by itself isolate language-specific transfer.
  • A direct test of this would be to evaluate a k-mer Jaccard classifier or a model fine-tuned on randomly shuffled PAWS-X labels on the same DNA pairs; comparable accuracy would undermine the transfer interpretation.
  • If the transfer is real, it suggests that structural sequence-similarity tasks share a common inductive bias across symbolic domains, so sentence-pair training could be a general pre-training recipe for genomic similarity search.
  • The observed failure on DNA-protein pairs may be too quick a conclusion: the negative examples in that task may be too easy or too hard relative to PAWS-X, so a task-construction effect, rather than the absence of transfer, could explain the random accuracy.
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 asks whether the capability transfer observed across natural languages in multilingual models also transfers from natural language to DNA/protein sequences. It constructs two biological sequence-pair classification tasks, DNA-pair similarity and DNA-protein-pair coding, modeled on the PAWS-X sentence-pair task. The authors fine-tune GPT-2 and BERT models on English PAWS-X data and evaluate on the biological tasks, reporting up to 78% accuracy for GPT-2-small and 89% for multilingual BERT on DNA-pair classification, while DNA-protein results remain near chance. They further study the influence of sequence length, training seed, pre-training data, and model scale, and include a word-vector analysis. The central claim is that experimental validation has confirmed that transfer from natural language to biological language is 'unequivocally present'.

Significance. If the central claim were established, the paper would offer a surprising and potentially useful result: that fine-tuning on a natural-language sentence-pair task can produce a model that performs DNA similarity judgments. The paper has useful strengths: the authors report open-source code and data, include a negative result for the DNA-protein task, and explore several factors (length, seed, pre-training, model scale) rather than only reporting a single accuracy number. However, the current evidence does not isolate a language-transfer mechanism. The task construction makes high accuracy achievable by trivial sequence-overlap heuristics, no such baseline is reported, and the headline accuracies are best-seed selections that overstate typical performance. These issues are load-bearing for the paper's main claim and require additional experiments and reporting before the conclusion can be accepted.

major comments (4)
  1. [§2.1, §3.1] The DNA-pair classification task is constructed so that positive pairs are BLAST homologs (E-value < 1e-5) and negative pairs are randomly selected sequences from other organisms. After truncation to 50 bp in §3.1, positive pairs will typically share long exact substrings while negative pairs share almost none. The reported accuracies of 78% (GPT-2) and 89% (BERT-multi) may therefore be explained by a model learning a string-overlap or k-mer-frequency heuristic from PAWS-X, rather than by transfer of a language-specific capability. No control baseline is reported: no direct DNA fine-tuning, no k-mer/Jaccard overlap classifier, no untrained model, no label-shuffled model. Please add such controls; if a non-transfer baseline also achieves high accuracy on the same test sets, the central claim is not supported.
  2. [§3.2, §3.3, Table 3] The paper states in §3.2 that 'each experiment was repeated multiple times, and the best-performing model from each set of experiments was selected.' Section 3.3 then reports that on DNA150 the seeds give mean accuracy 0.636, median 0.624, standard deviation 0.082, with a minimum of 0.52 and a maximum of 0.79. The numbers in Tables 1, 3, 4, and 6 are therefore best-of-selection results, not typical results. The gap between the best (0.79) and the mean (0.636) is large, and the minimum is close to chance. Please report mean, standard deviation, and number of seeds for every DNA result, and state whether the mean accuracy is significantly above the no-transfer baseline. Without this, the headline 78%/89% figures are not robust evidence for transfer.
  3. [Abstract, Conclusion] The abstract and conclusion claim that 'Experimental validation has confirmed that the transfer of capabilities from natural language to biological language is unequivocally present.' Given the missing control baselines and the seed-sensitivity documented in §3.3, this is an overclaim. The evidence at most shows that a PAWS-X-fine-tuned transformer can score above chance on a BLAST-based sequence-similarity task, not that the underlying capability is linguistic. In addition, the conclusion contains the sentence 'can achieve an accuracy of 829%' which is a typographical error for 89% or 82%; this must be corrected.
  4. [§3.7] The word-vector analysis reports that the mean nearest neighbor distance between DNA and English word vectors decreases from 50.09 to 29.79 after fine-tuning, and the PCA figures show increased overlap. This is a post-hoc analysis on the same fine-tuned model and is presented without a baseline or statistical test. For example, a random embedding or a model fine-tuned on a non-similarity task could exhibit a similar distance change. If this analysis is retained as supporting evidence, add appropriate controls and significance testing.
minor comments (5)
  1. [General] Several figures are referenced in the text (Figs. 2, 3, 4, 5, 7, 8, 9) but no figure images appear in the manuscript text provided; please ensure all figures are actually included and legible.
  2. [§3.1] The sentence 'there is no information loss due to truncation' is misleading: truncating sequences to the first 50 bp discards the rest of the original sequences, even if it brings the input length within the model's context window. Please rephrase.
  3. [Table 3] The Dna150s dataset contains 30% completely identical pairs and 40% containment pairs by construction, so the very high accuracies on that dataset are expected and should not be conflated with the harder Dna150 results. Consider presenting Dna150s and Dna150 separately in the discussion.
  4. [References] Reference formatting is inconsistent: reference [24] is given as 'paper github' with a URL, and the arXiv identifiers in references [13] and [16] appear to be identical. Please standardize the bibliography.
  5. [Abstract] The abstract uses 'precision reached 89%' where the rest of the paper uses 'accuracy'; please make the terminology consistent.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper reports empirical transfer results; no fitted parameter is renamed as a prediction and no load-bearing self-citation chain is present.

full rationale

This paper is an empirical study, not a derivation, so the circularity patterns defined here do not apply in their usual form. The central claim is that fine-tuning GPT-2 or BERT on the English PAWS-X sentence-pair dataset yields high accuracy on a separately constructed DNA-pair similarity task. The DNA labels are not defined by the model's output or by any parameter fitted to the DNA test set: positive pairs are BLAST homologs (E < 1e-5) and negative pairs are randomly selected sequences from other organisms, all constructed before model training. The 78% and 89% accuracies are measured outcomes, not quantities derived by construction from the fine-tuning data. Section 3.7's word-vector analysis uses the same fine-tuned model to visualize distances between DNA and English word vectors, but this is a post-hoc explanatory visualization, not the definition of the accuracy metric, and it does not make the reported transfer accuracy true by definition. The paper contains no uniqueness theorem imported from the authors' prior work, no ansatz smuggled in via self-citation, and no renamed known result. The absence of control baselines such as direct DNA fine-tuning or k-mer overlap classifiers, and the seed-selection practice in Section 3.3, are experimental-design and evidence-quality concerns rather than circularity. Accordingly, the circularity score is 0.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claim rests on several hand-set dataset construction thresholds, which function as free parameters, and on two domain assumptions about the analogy between sentence similarity and DNA similarity and about what the accuracy metric measures. No genuinely new entities are introduced. The threshold choices strongly influence the reported accuracies, and the absence of a control baseline means the assumptions are not independently validated.

free parameters (5)
  • BLAST E-value threshold for positive DNA pairs = < 1e-5
    Chosen by hand to define which DNA pairs count as similar; directly determines task difficulty in Section 2.1.
  • Sequence similarity score cutoffs = > 0.85 for similar, < 0.55 for dissimilar
    Hand-set thresholds in Section 3.2 for constructing the DNA150 and DNA50 datasets.
  • Length difference limit = no more than 10%
    Hand-set criterion for similar and dissimilar pairs in Section 3.2.
  • DNA sequence truncation length = 50 base pairs
    Chosen to match PAWS-X token lengths after initial raw sequences were too long; affects all headline accuracy numbers.
  • Composition of Dna150s dataset = 30% identical pairs, 40% containment pairs
    Hand-specified mixture in Table 2 that makes the simple dataset easier; no justification beyond experimental design.
assumptions (3)
  • domain assumption Sentence-pair similarity judgment in natural language is structurally analogous to DNA sequence similarity judgment.
    This analogy is the entire basis for using PAWS-X as the source task; it is asserted in Sections 1 and 2.1, not tested.
  • domain assumption A unified tokenizer (BPE/WordPiece) is necessary and sufficient to enable capability transfer between languages and DNA.
    Section 2.2 states this based on prior multilingual work; the paper does not ablate tokenizer choice.
  • domain assumption Classification accuracy on the constructed DNA-pair dataset measures 'language capability transfer'.
    Section 3.1 interprets high accuracy as evidence of transfer, but without a control baseline the metric could reflect token-overlap heuristics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Can linguists better understand DNA?." pith.science (2026). https://pith.science/paper/5YB5SSCS

@misc{pith2026241207678,
  author       = {Pith},
  title        = {Pith review of: Can linguists better understand DNA?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5YB5SSCS}},
  note         = {Machine review of arXiv:2412.07678}
}
read the original abstract

Multilingual transfer ability, which reflects how well models fine-tuned on one source language can be applied to other languages, has been well studied in multilingual pre-trained models. However, the existence of such capability transfer between natural language and gene sequences/languages remains under explored.This study addresses this gap by drawing inspiration from the sentence-pair classification task used for evaluating sentence similarity in natural language. We constructed two analogous tasks: DNA-pair classification(DNA sequence similarity) and DNA-protein-pair classification(gene coding determination). These tasks were designed to validate the transferability of capabilities from natural language to gene sequences. Even a small-scale pre-trained model like GPT-2-small, which was pre-trained on English, achieved an accuracy of 78% on the DNA-pair classification task after being fine-tuned on English sentence-pair classification data(XTREME PAWS-X). While training a BERT model on multilingual text, the precision reached 89%. On the more complex DNA-protein-pair classification task, however, the model's output was barely distinguishable from random output.Experimental validation has confirmed that the transfer of capabilities from natural language to biological language is unequivocally present. Building on this foundation, we have also investigated the impact of model parameter scale and pre-training on this capability transfer. We provide recommendations for facilitating the transfer of capabilities from natural language to genetic language,as well as new approaches for conducting biological research based on this capability.This study offers an intriguing new perspective on exploring the relationship between natural language and genetic language.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Find Central Dogma Again: Leveraging Multilingual Transfer in Large Language Models

    q-bio.GN 2025-02 reject novelty 5.0 of 10

    A GPT-2 model fine-tuned on multilingual sentence similarity achieves at best 81% accuracy on classifying matching vs non-matching DNA-protein pairs, but the result is highly seed-dependent and only with an easy test set.

  2. Human Genome Book: Words, Sentences and Paragraphs

    q-bio.OT 2025-01 reject novelty 4.0 of 10

    A GPT-2 model trained on DNA, protein, and English is used to segment the human genome into book-like words, sentences, paragraphs, and chapters, but transfer to DNA is only hypothesized, not validated.

Reference graph

Works this paper leans on

34 extracted references · 13 canonical work pages · cited by 2 Pith papers

  1. [1]

     The input format was sequence1 + sequence2, and the output was the corresponding label

    Fine-tuning GPT2 on PAWS-X English Data:  We fine-tuned the GPT2-small model on the PAWS-X English dataset.  The input format was sequence1 + sequence2, and the output was the corresponding label

    work page

  2. [2]

    Evaluation on Biosequence Similarity Tasks:  After fine-tuning, we evaluated the model on the DNA-protein pair similarity judgment task and the DNA-DNA sequence pair similarity judgment task

    work page

  3. [3]

    Tokenizer Considerations Given the importance of tokenization in language capability transfer, we ensured consistency by using:  GPT2 Models: Utilized the BPE tokenizer

    Testing Other Model Variants:  We extended our experiments to include other parameter scales of GPT2 models and BERT models to compare performance. Tokenizer Considerations Given the importance of tokenization in language capability transfer, we ensured consistency by using:  GPT2 Models: Utilized the BPE tokenizer.  BERT Models: Used the WordPiece tok...

    work page 1964

  4. [4]

    The input is specific words, and we use the last hidden layer of the GPT-2 network, applying average pooling to obtain a vector representation for each word

    Word Vector Generation: Using the entire set of English words and DNA vocabulary, we generate word vectors. The input is specific words, and we use the last hidden layer of the GPT-2 network, applying average pooling to obtain a vector representation for each word. Each word is represented as a 768-dimensional vector

    work page

  5. [5]

    closeness

    Measuring Similarity Between DNA Word Vectors and English Word Vectors: We use the Mean Nearest Neighbor Distance (MNND) method to calculate the similarity. This method computes the distance from each vector in one group to its nearest neighbor in the other group and then takes the average of these distances. This allows us to measure the "closeness" betw...

    work page

  6. [6]

    Pre-train Large Models on Both Natural Language and Biological Sequences: Develop large models that are pre-trained on both natural language and biological sequence data to capture the nuances of both domains

    work page

  7. [7]

    For example, this study used the PAWS-X text similarity dataset as a basis to construct a biological sequence similarity pair dataset

    Construct a Verifiable Test Dataset for Natural Language to Biological Sequence Transfer: Create a dataset specifically designed to test the transferability from natural language to biological sequences. For example, this study used the PAWS-X text similarity dataset as a basis to construct a biological sequence similarity pair dataset

    work page

  8. [8]

    Fine-tune Models Based on the Above Dataset: Fine-tune models using the newly constructed dataset and select the best-performing model as the foundational multimodal model linking natural language to biological sequences

    work page

Show all 34 references
  1. [9]

    foundational multimodal model

    Fine-tune the Foundational Multimodal Model on Other Natural Language Tasks: Further fine-tune the selected "foundational multimodal model" on different types of natural language tasks. For instance, use an English text summarization dataset to fine-tune a new model

  2. [10]

    This step explores whether the summarization capability can be effectively transferred from natural language to biological sequences

    Process Biological Sequences : Apply the fine-tuned text summarization model to process long DNA sequences, generating shorter DNA-based summaries. This step explores whether the summarization capability can be effectively transferred from natural language to biological sequences

  3. [11]

    Biological Experiment Verification: Conduct specific biological experiments to validate the results from step 5. Using text summarization as an example, generate summary DNA sequences from multiple long DNA sequences, then use BLAST (Basic Local Alignment Search Tool) to index...

  4. [12]

    Zhou, et al., Dnabert-2: Efficient foundation model and benchmark for multi-species

    Z. Zhou, et al., Dnabert-2: Efficient foundation model and benchmark for multi-species

  5. [13]

    genome, arXiv preprint arXiv:2306.15006 (2023)

    2023 arXiv

  6. [14]

    Nguyen, et al., Hyenadna: Long-range genomic sequence modeling at single nucleotide resolution, Advances in neural information processing systems 36 (2024)

    E. Nguyen, et al., Hyenadna: Long-range genomic sequence modeling at single nucleotide resolution, Advances in neural information processing systems 36 (2024)

    2024

  7. [15]

    Yang, et al., scbert as a large-scale pretrained deep language model for cell type annotation of single-cell rna-seq data, Nature Machine Intelligence 4, 852–866 (2022)

    F. Yang, et al., scbert as a large-scale pretrained deep language model for cell type annotation of single-cell rna-seq data, Nature Machine Intelligence 4, 852–866 (2022)

    2022

  8. [16]

    A. Elnaggar, et al., Prottrans: Toward understanding the language of life through selfsupervised learning, IEEE transactions on pattern analysis and machine intelligence 44, 7112–7127 (2021)

    2021

  9. [17]

    Brandes, D

    N. Brandes, D. Ofer, Y. Peleg, N. Rappoport, M. Linial, Proteinbert: a universal deeplearning model of protein sequence and function, Bioinformatics 38, 2102–2110 (2022)

    2022

  10. [18]

    Lin, et al., Evolutionary-scale prediction of atomic-level protein structure with a language model, Science 379, 1123–1130 (2023)

    Z. Lin, et al., Evolutionary-scale prediction of atomic-level protein structure with a language model, Science 379, 1123–1130 (2023)

    2023

  11. [19]

    & Davuluri, R

    Ji, Y., Zhou, Z., Liu, H. & Davuluri, R. V. DNABERT: Pre-trained bidirectional encoder representations from transformers model for DNA-language in genome. Bioinformatics 37, 2112–2120 (2021). doi:doi.org/10.1093/bioinformatics/btab083

    2021 doi

  12. [20]

    doi: https://doi.org/10.1101/2023.06.12.544594

    Veniamin Fishman, et al.GENA-LM: A Family of Open-Source Foundational DNA Language Models for Long Sequences.BioRxiv 2023.06.12.544594. doi: https://doi.org/10.1101/2023.06.12.544594

    2023 doi

  13. [21]

    BioRxiv 2023.07.11.548628; doi: https://doi.org/10.1101/2023.07.11.548628

    Daoan Zhang, et al.DNAGPT: A Generalized Pretrained Tool for Multiple DNA Sequence Analysis Tasks. BioRxiv 2023.07.11.548628; doi: https://doi.org/10.1101/2023.07.11.548628

    2023 doi

  14. [22]

    Wang, B. et al. Pre-trained Language Models in Biomedical Domain: A Systematic Survey. ACM Comput. Surv.56, 1–52 (2023).doi:doi.org/10.1145/3611651

    2023 doi

  15. [23]

    & Gao, G

    Chen, Z., Wei, L. & Gao, G. Foundation models for bioinformatics. Quant. Biol. (2024) doi:10.1002/qub2.69

    2024 doi

  16. [25]

    Fei Yuan,etc.How Vocabulary Sharing Facilitates Multilingualism in LLaMA?arXiv (2024) doi:https://doi.org/10.48550/arXiv.2306.06688

    2024 doi arXiv

  17. [26]

    Alexis Conneau, Shijie Wu, Haoran Li, Luke Zettlemoyer, and Veselin Stoyanov. 2020. Emerging Cross-lingual Structure in Pretrained Language Models. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6022–6034, Online. Association ...

    2020

  18. [27]

    Specialists: An Empirical Revisiting on Multilingual Transfer Ability

    Jiacheng Ye, Xijia Tao, Lingpeng Kong.Language Versatilists vs. Specialists: An Empirical Revisiting on Multilingual Transfer Ability. arXiv (2024) doi:10.48550/arxiv.2407.11435

    2024 doi arXiv

  19. [28]

    Hetong Wang, Pasquale Minervini, and Edoardo Ponti. 2024. Probing the Emergence of Cross-lingual Alignment during LLM Training. In Findings of the Association for Computational Linguistics: ACL 2024, pages 12159–12173, Bangkok, Thailand. Association for Computational Linguistics

    2024

  20. [29]

    arXiv(2020) https://arxiv.org/abs/2003.11080

    Junjie Hu, etc.XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization. arXiv(2020) https://arxiv.org/abs/2003.11080

    2020 arXiv

  21. [30]

    BioRxiv 2024.07.01.600583; doi: https://doi.org/10.1101/2024.07.01.600583

    Thomas Hayes, et al.Simulating 500 million years of evolution with a language model. BioRxiv 2024.07.01.600583; doi: https://doi.org/10.1101/2024.07.01.600583

    2024 doi

  22. [31]

    LucaOne: Generalized Biological Foundation Model with Unified Nucleic Acid and Protein Language

    Yong He, et al. LucaOne: Generalized Biological Foundation Model with Unified Nucleic Acid and Protein Language. BioRxiv 2024.05.10.592927; doi: https://doi.org/10.1101/2024.05.10.592927

    2024 doi

  23. [32]

    Zhang, K., Zhou, R., Adhikarla, E. et al. A generalist vision–language foundation model for diverse biomedical tasks. Nat Med (2024). https://doi.org/10.1038/s41591-024-03185-2

    2024 doi

  24. [33]

    Minghao Xu, Xinyu Yuan, Santiago Miret, and Jian Tang. 2023. ProtST: multi-modality learning of protein sequences and biomedical texts. In Proceedings of the 40th International Conference on Machine Learning (ICML'23), Vol. 202. JMLR.org, Article 1615, 38749–38767

    2023

  25. [34]

    Zeyuan Wang, etc. 2024. InstructProtein: Aligning Human and Protein Language via Knowledge Instruction. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1114–1136, Bangkok, Thailand. Association for Compu...

    2024

  26. [35]

    paper github, https://github.com/maris205/find_central-dogma_again

Pith tools

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