Tasks SotA Datasets Papers Methods Submit About

Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable Benchmarks All SotA Datasets Papers Methods

Community

Submit Results About

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Models/MolXPT

MolXPT

Reported on 31 benchmarks across 5 tasks · 1 paper · 5 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Medical27 results

Molecule CaptioningonChEBI-20
Text2Mol· 2023-05-18
59.4
best: 60.3 (BioT5)
SOTA
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Atomistic DescriptiononSIDER
ROC-AUC· 2023-05-18
71.7
best: 91.11 (BioAct-Het)
SOTA
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Atomistic DescriptiononBACE
ROC-AUC· 2023-05-18
88.4
SOTA
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
Exact Match· 2023-05-18
21.5
best: 53.3 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
Frechet ChemNet Distance (FCD)· 2023-05-18
0.45
best: 2.49 (MolT5-small)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
MACCS FTS· 2023-05-18
85.9
best: 97.3 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
Morgan FTS· 2023-05-18
66.7
best: 93.1 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
Parameter Count· 2023-05-18
350000000
best: 770000000 (MolT5-Large)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
RDK FTS· 2023-05-18
75.7
best: 95 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
Text2Mol· 2023-05-18
57.8
best: 59.3 (MolReGPT (GPT-4-0413))
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Drug DiscoveryonChEBI-20
Validity· 2023-05-18
98.3
best: 100 (BioT5+)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecule CaptioningonChEBI-20
BLEU-2· 2023-05-18
59.4
best: 73.2 (Mol-LLM (Mistral-Instruct-v0.2))
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecule CaptioningonChEBI-20
BLEU-4· 2023-05-18
50.5
best: 60.8 (MolReFlect)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecule CaptioningonChEBI-20
METEOR· 2023-05-18
62.6
best: 68.1 (BioT5+)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecule CaptioningonChEBI-20
ROUGE-1· 2023-05-18
66
best: 71 (BioT5+)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecule CaptioningonChEBI-20
ROUGE-2· 2023-05-18
51.1
best: 61.8 (XMolCap)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecule CaptioningonChEBI-20
ROUGE-L· 2023-05-18
59.7
best: 65.3 (PEIT-GEN)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Atomistic DescriptiononHIV dataset
AUC· 2023-05-18
0.781
best: 0.81 (DVMP)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Atomistic DescriptiononTox21
ROC-AUC· 2023-05-18
77.1
best: 92.4 (Deep-CBN)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
Exact Match· 2023-05-18
21.5
best: 53.3 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
Frechet ChemNet Distance (FCD)· 2023-05-18
0.45
best: 2.49 (MolT5-small)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
MACCS FTS· 2023-05-18
85.9
best: 97.3 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
Morgan FTS· 2023-05-18
66.7
best: 93.1 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
Parameter Count· 2023-05-18
350000000
best: 770000000 (MolT5-Large)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
RDK FTS· 2023-05-18
75.7
best: 95 (LDMol)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
Text2Mol· 2023-05-18
57.8
best: 59.3 (MolReGPT (GPT-4-0413))
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Text-based de novo Molecule GenerationonChEBI-20
Validity· 2023-05-18
98.3
best: 100 (BioT5+)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688

Methodology4 results

Molecular Property PredictiononSIDER
ROC-AUC· 2023-05-18
71.7
best: 91.11 (BioAct-Het)
SOTA
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecular Property PredictiononBACE
ROC-AUC· 2023-05-18
88.4
SOTA
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecular Property PredictiononHIV dataset
AUC· 2023-05-18
0.781
best: 0.81 (DVMP)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688
Molecular Property PredictiononTox21
ROC-AUC· 2023-05-18
77.1
best: 92.4 (Deep-CBN)
MolXPT: Wrapping Molecules with Text for Generative Pre-training arXiv:2305.10688