List your company

Text-guided molecule generation on ChEBI-20 (test)

100MACCS FTS SimilarityGround Truth
21.27241.71162.1582.589Apr 25, 2022Oct 4, 2022Mar 15, 2023Aug 25, 2023Feb 3, 2024Jul 14, 2024Dec 24, 2024

Evaluation results

MethodLinks
Ground Truth2024.021001001000100100060.9100
Ground Truth2022.041001001000100100060.9100
LDMolFamily=DM2024.0597.392.6536.759593.10.2—94.1
bioT5+Family=AR2024.0590.787.252.212.77683.577.90.35—100
MolReGPTbackbone=GPT-4-0314, shots=10-shot2023.0690.385.72817.1480.573.90.4159.389.9
bioT5Family=AR2024.0588.686.741.315.09780.173.40.43—100
PEIT-LLM-LLaMa3.1#Params=8B2024.1287.583.6—18.0377.666.1——97
TGM-DLMParameters=~180M, Correction Phase=Disabled2024.0287.482.824.216.89777.172.20.8958.978.9
MolXPTNumber of Parameters=350M2023.0585.9—21.5—75.766.70.4557.898.3
TGM-DLMParameters=~180M, Correction Phase=Enabled2024.0285.482.624.217.00373.968.80.7758.187.1
TGM-DLMFamily=DM2024.0585.482.624.217.00373.968.80.77—87.1
MolReGPTbackbone=GPT-3.5-turbo, shots=10-shot2023.0684.77913.924.9170.862.40.5757.188.7
MolReGPT#Params=N/A2024.1284.779—24.9170.862.4——88.7
MolT5Model Size=Large2022.0483.485.431.116.07174.668.41.255.490.5
MolT5-largeNumber of Parameters=800M2023.0583.4—31.1—74.668.41.255.490.5
MolT5-largemodel_size=large2023.0683.485.431.116.07174.668.41.255.490.5
MolT5largeFamily=AR2024.0583.485.431.116.07174.668.41.2—90.5
PEIT-LLM-Qwen2.5#Params=7B2024.1283.281—21.13373.561.9——95
T5Model Size=Large2022.0482.385.427.916.72173.1671.2255.290.2
T5-largemodel_size=large2023.0682.385.427.916.72173.1671.2255.290.2
T5largeFamily=AR2024.0582.385.427.916.72173.1671.22—90.2
MolReGPTbackbone=Llama-2-7B, shots=2-shot2023.0680.869.32.236.7771.760.94.914.976.1
MolXPTFamily=AR2024.0575.785.945—66.721.5——98.3
GIT-MolFamily=AR2024.0573.875.65.126.31558.251.9——92.8
T5-BaseParameters=~220M, Pre-training=C4 dataset2024.0273.176.26.924.9560.554.52.4849.966
T5Model Size=Base2022.0473.176.26.924.9560.554.52.4849.966
T5-basemodel_size=base2023.0673.176.26.924.9560.554.52.4849.966
T5baseFamily=AR2024.0573.176.56.924.9560.554.52.48—66
MolT5-BaseParameters=~220M, Pre-training=C4 and ZINC-152024.0272.176.98.124.45858.852.92.1849.677.2
MolT5Model Size=Base2022.0472.176.98.124.45858.852.92.1849.677.2
MolT5-mediumNumber of Parameters=250M2023.0572.1—8.1—58.852.92.1849.677.2
MolT5-basemodel_size=base2023.0672.176.98.124.45858.852.92.1849.677.2
MolT5baseFamily=AR2024.0572.176.98.124.45858.852.92.18—77.2
GPT-3.5-turbomode=zero-shot2023.0670.548.91.952.1346.236.72.0547.980.2
GPT-3.5-turbo#Params=N/A2024.1270.548.9—52.1346.236.7——80.2
T5Model Size=Small2022.0470.474.16.427.70357.852.52.8947.960.8
MolT5Model Size=Small2022.0470.375.57.925.98856.851.72.4948.272.1
MolT5-smallNumber of Parameters=77M2023.0570.3—7.9—56.851.72.4948.272.1
RNN2022.0459.165.20.538.094036.24.5540.954.2
TransformerArchitecture=Vanilla Transformer, Layers=6 encoder/decoder2024.024849.9057.663221.711.3227.790.6
Transformer2022.044849.9057.663221.711.3227.790.6
Transformer2023.064849.9057.663221.711.3227.790.6
TransformerFamily=AR2024.054849.9057.663221.711.32—90.6
Mol-Instructions#Params=8B2024.1241.234.5—41.36723.114.7——100
LLaMa3.1#Params=8B2024.1241.127—43.18324.813.8——36.8
Qwen2.5#Params=7B2024.1240.321.7—50.5527.611——24.5
LLaMa3#Params=7B2024.1237.226.1—45.78821.312.7——33
Llama-2-7Bmode=zero-shot2023.0624.310.4084.1811.98.942.0114.863.1