Browse › Medical › Text-based de novo Molecule Generation › ChEBI-20

Text-based de novo Molecule Generation archive 2025-07-28

ChEBI-20 Benchmark (Text-based de novo Molecule Generation)

20 rows 19 with code listed 10 metrics Dataset page

Text-based de novo molecule generation involves utilizing natural language processing (NLP) techniques and chemical information to generate entirely new molecular structures. In this approach, molecular structures are typically encoded as text strings, resembling chemical formulas or SMILES (Simplified Molecular Input Line Entry System). Subsequently, by applying NLP models such as recurrent neural networks (RNNs) or Transformer models, these text strings are processed to generate novel molecular structures with desired properties.

The archive carries no text for this table; the description above is the archive's text for the task Text-based de novo Molecule Generation. archive 2025-07-28

Over time archive 2025-07-28

The chart needs JavaScript; the table below carries every value.

Direction inferred from the metric name, not from the archive: BLEU (higher is better), Frechet ChemNet Distance (FCD) (lower is better). Not inferred (points only, no best-so-far line): Exact Match, Levenshtein, MACCS FTS, Morgan FTS, RDK FTS, Text2Mol, Validity, Parameter Count. Points are placed at the row's paper date; 20 of 20 rows carry one.

Results archive 2025-07-28

Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.

Paper Code Ran Syntology Report
1 LDMol 92.653.30.206.75097.393.195.094.1 – Paper Code 2024 6 of 9 ran · 3 unverified report
2 MolReFlect 90.351.011.8492.981.386.097.7 – Paper – 2024 no code linked report
3 BioT5+ 87.252.20.35312.77690.777.983.557.9100252000000 – Paper Code 2024 linked, not harvested report
4 BioT5 86.741.3.4315.09788.673.480.157.6100252000000 – Paper Code 2023 1 of 1 ran · 0 unverified report
5 MolReGPT (GPT-4-0413) 85.728.00.4117.1490.373.980.559.389.9None – Paper Code 2023 linked, not harvested report
6 MolT5-Large 85.430.21.2016.0783.468.474.655.490.5770000000 – Paper Code 2022 5 of 5 ran · 0 unverified report
7 Text+Chem T5-augm base 85.332.2.0516.8790.175.781.694.3220000000 – Paper Code 2023 linked, not harvested report
8 TGM-DLM w/o corr 82.824.20.8916.89787.472.277.158.978.9180000000 – Paper Code 2024 14 of 21 ran · 7 unverified report
9 TGM-DLM 82.624.20.7717.00385.468.873.958.187.1180000000 – Paper Code 2024 14 of 21 ran · 7 unverified report
10 MolFM-Base 82.221.019.44585.475.869.758.389.2296200000 – Paper Code 2023 linked, not harvested report
11 Text+Chem T5-augm small 81.519.10.0621.7886.467.274.495.160000000 – Paper Code 2023 linked, not harvested report
12 MolT5-Large-HV 81.031.40.4416.75887.272.278.659.099.6770000000 – Paper Code 2022 5 of 5 ran · 0 unverified report
13 MolFM-Small 80.316.920.86883.472.166.257.385.913620000 – Paper Code 2023 linked, not harvested report
14 MolReGPT (GPT-3.5-turbo) 79.013.90.5724.9184.762.470.857.188.7 – Paper Code 2023 linked, not harvested report
15 MolT5-base 76.98.12.1824.45872.152.958.849.677.2220000000 – Paper Code 2022 5 of 5 ran · 0 unverified report
16 GIT-Mol-caption 75.65.126.31573.851.958.292.8 – Paper Code 2023 1 of 1 ran · 0 unverified report
17 MolT5-small 75.57.92.4925.98870.351.756.848.272.160000000 – Paper Code 2022 5 of 5 ran · 0 unverified report
18 Text+Chem T5 base 7521.20.06127.3987.469.776.779.2220000000 – Paper Code 2023 linked, not harvested report
19 Text+Chem T5 small 73.915.70.06628.5485.96673.677.660000000 – Paper Code 2023 linked, not harvested report
20 MolXPT 21.50.4585.966.775.757.898.3350000000 – Paper Code 2023 linked, not harvested report

All 20 rows shown. 20 link to a paper page on this site; 0 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28

Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 9 rows have a graph line, from 5 distinct papers; 9 rows (5 papers) have at least one sample that ran. Counting each paper once: Syntology ran 27 of 37 samples; 10 unverified. Separately, 22 of those 37 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-25. Per-sample status is on the paper page.

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