Retrosynthesis on USPTO-50k Reaction type known (test) benchmark leaderboard · SOTA2 Research

Retrosynthesis on USPTO-50k Reaction type known (test)

67.2Top-1 AccuracyNAG2G

Evaluation Results

Method	Links
NAG2G2023.09		67.2	86.4	90.5	93.8
RetroKNN2024.03		66.7	88.2	93.6	96.6
MARS2023.09		66.2	85.8	90.2	92.9
UAlign2024.03		66.2	86.9	91.9	95.1
MARS2022.09		66.2	85.8	90.2	92.9
RetroComposer2023.09		65.9	85.8	89.5	91.5
SemiRetro2023.09		65.8	85.7	89.8	92.8
Dual (Ours)2020.07		65.7	81.9	84.7	85.9
DualTF2023.09		65.7	81.9	84.7	85.9
DualTF2022.09		65.7	81.9	84.7	85.9
RetroPrime2023.09		64.8	81.6	85	86.9
RetroPrime2024.03		64.8	81.6	85	86.9
GLN2020.07		64.2	79.1	85.2	90
GLN2023.09		64.2	79.1	85.2	90
GLN2024.03		64.2	79.1	85.2	90
GLN2022.09		64.2	79.1	85.2	90
Retroformer2023.09		64	82.5	86.7	90.2
Retroformer2024.03		64	82.5	86.7	90.2
GraphRETRO2020.07		63.9	81.5	85.2	88.1
GraphRetro2023.09		63.9	81.5	85.2	88.1
LocalRetro2024.03		63.9	86.8	92.4	96.3
GraphRetro2024.03		63.9	81.5	85.2	88.1
GraphRetro2022.09		63.9	81.5	85.2	88.1
G2Retro2023.09		63.6	83.6	88.4	91.5
RetroXpert2020.07		62.1	75.8	78.5	80.9
RetroXpert2023.09		62.1	75.8	78.5	80.9
RetroXpert2024.03		62.1	75.8	78.5	80.9
RetroXpert2022.09		62.1	75.8	78.5	80.9
G2Gs2020.07		61	81.3	86	88.7
G2G2023.09		61	81.3	86	88.7
G2G2024.03		61	81.3	86	88.7
G2Gs2022.09		61	81.3	86	88.7
MEGAN2023.09		60.7	82	87.5	91.6
MEGAN2024.03		60.7	82	87.5	91.6
MEGAN2022.09		60.7	82	87.5	91.6
Transformer2020.07		59	74.8	78.1	81.1
SCROP2023.09		59	74.8	78.1	81.1
SCROP2024.03		59	74.8	78.1	81.1
SCROP2022.09		59	74.8	78.1	81.1
GET2023.09		57.4	71.3	74.8	77.4
Transformer2024.03		57.1	71.5	75	77.7
NeuralSym2020.07		55.3	76	81.4	85.1
NeuralSym2023.09		55.3	76	81.4	85.1
NeuralSym2024.03		55.3	76	81.4	85.1
NeuralSym2022.09		55.3	76	81.4	85.1
retrosim2020.07		52.9	73.8	81.2	88.1
RetroSim2023.09		52.9	73.8	81.2	88.1
RetroSim2024.03		52.9	73.8	81.2	88.1
RetroSim2022.09		52.9	73.8	81.2	88.1
LSTM2020.07		37.4	52.4	57	61.7