Papers › Translation between Molecules and Natural Language

Translation between Molecules and Natural Language

25 Apr 2022arXiv:2204.11817archive 2025-07-28

Carl Edwards, Tuan Lai, Kevin Ros, Garrett Honke, Kyunghyun Cho, Heng Ji

We present MolT5 - a self-supervised learning framework for pretraining models on a vast amount of unlabeled natural language text and molecule strings. MolT5 allows for new, useful, and challenging analogs of traditional vision-language tasks, such as molecule captioning and text-based de novo molecule generation (altogether: translation between molecules and language), which we explore for the first time. Since MolT5 pretrains models on single-modal data, it helps overcome the chemistry domain shortcoming of data scarcity. Furthermore, we consider several metrics, including a new cross-modal embedding-based metric, to evaluate the tasks of molecule captioning and text-based molecule generation. Our results show that MolT5-based models are able to generate outputs, both molecules and captions, which in many cases are high quality.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 5 code samples harvested from 1 repository linked to this paper: 5 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 0 violated, 5 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 5 samples from 1 repository, 5 ran (5 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

blender-nlp/MolT5 officialmentioned in papermentioned on GitHubpytorchBSD-3-ClauseSyntology: 5 samples, 5 ran (0 instrument-failed) · 0 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

5 samples harvested: 5 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 0 violated, 5 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

5ran

Licence: 0 of the 5 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from blender-nlp/MolT5. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

blender-nlp/MolT5 official repository 5 samples, 5 ran (0 instrument-failed) · 0 unverified
build_smiles_vocab baselines/main_RNN_caption2smiles.py ran BSD-3-Clause (permissive) · c4e38096a64d0bfa · report fixture: signature dicts:list/1/dict · returned str · level recorded 2026-09-24
build_smiles_vocab baselines/main_transformer_caption2smiles.py ran BSD-3-Clause (permissive) · 5567d53edbc7d0d2 · report fixture: signature dicts:list/1/dict · returned str · level recorded 2026-09-24
create_mask in 35 papers baselines/main_transformer_caption2smiles.py ran BSD-3-Clause (permissive) · c8ed36180a390809 · report fixture: signature src:Tensor/2/int, tgt:Tensor/2/int · returned tuple · level recorded 2026-09-24
generate_square_subsequent_mask in 30 papers baselines/main_transformer_caption2smiles.py ran BSD-3-Clause (permissive) · 5da85d40a7a21ba4 · report fixture: signature sz:int/0/int · returned Tensor · level recorded 2026-09-24
maskNLLLoss baselines/main_RNN_caption2smiles.py ran BSD-3-Clause (permissive) · 25307394e75462fd · report fixture: signature output:Tensor/2/float, target:Tensor/1/int, mask:Tensor/1/float · returned tuple · level recorded 2026-09-24

Tasks

Drug DiscoveryMolecule CaptioningSelf-Supervised LearningText-based de novo Molecule GenerationTranslation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Molecule Captioning ChEBI-20 MolT5-Large BLEU-2 59.4 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Large BLEU-4 50.8 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Large METEOR 61.4 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Large ROUGE-1 65.4 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Large ROUGE-2 51.0 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Large ROUGE-L 59.4 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Large Text2Mol 58.2 #15 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base BLEU-2 54.0 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base BLEU-4 45.7 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base METEOR 56.9 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base ROUGE-1 63.4 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base ROUGE-2 48.5 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base ROUGE-L 57.8 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Base Text2Mol 54.7 #28 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small BLEU-2 51.9 #31 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small BLEU-4 43.6 #31 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small METEOR 55.1 #31 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small ROUGE-1 62.0 #31 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small ROUGE-2 46.9 #31 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small ROUGE-L 56.3 #31 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 MolT5-Small Text2Mol 54.0 #31 of 33 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Large BLEU-2 76.9 #3 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Large BLEU-4 55.6 #3 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Large METEOR 74.3 #3 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Large ROUGE-1 77.7 #3 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Large ROUGE-2 58.0 #3 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Large ROUGE-L 55.7 #3 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Base BLEU-2 73.8 #5 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Base BLEU-4 53.5 #5 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Base METEOR 71.8 #5 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Base ROUGE-1 75.0 #5 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Base ROUGE-2 55.9 #5 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Base ROUGE-L 53.9 #5 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Small BLEU-2 70.9 #6 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Small BLEU-4 51.2 #6 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Small METEOR 70.1 #6 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Small ROUGE-1 74.5 #6 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Small ROUGE-2 55.8 #6 of 6 Archive leaderboard report
Molecule Captioning L+M-24 MolT5-Small ROUGE-L 54.4 #6 of 6 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large BLEU 85.4 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Exact Match 30.2 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Frechet ChemNet Distance (FCD) 1.20 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Levenshtein 16.07 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large MACCS FTS 83.4 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Morgan FTS 68.4 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Parameter Count 770000000 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large RDK FTS 74.6 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Text2Mol 55.4 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large Validity 90.5 #6 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV BLEU 81.0 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Exact Match 31.4 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Frechet ChemNet Distance (FCD) 0.44 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Levenshtein 16.758 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV MACCS FTS 87.2 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Morgan FTS 72.2 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Parameter Count 770000000 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV RDK FTS 78.6 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Text2Mol 59.0 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-Large-HV Validity 99.6 #12 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base BLEU 76.9 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Exact Match 8.1 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Frechet ChemNet Distance (FCD) 2.18 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Levenshtein 24.458 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base MACCS FTS 72.1 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Morgan FTS 52.9 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Parameter Count 220000000 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base RDK FTS 58.8 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Text2Mol 49.6 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-base Validity 77.2 #15 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small BLEU 75.5 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Exact Match 7.9 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Frechet ChemNet Distance (FCD) 2.49 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Levenshtein 25.988 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small MACCS FTS 70.3 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Morgan FTS 51.7 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Parameter Count 60000000 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small RDK FTS 56.8 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Text2Mol 48.2 #17 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 MolT5-small Validity 72.1 #17 of 20 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections