A deep learning framework for accurate reaction prediction and its application on high-throughput experimentation data

Table 2 Comparison of model performance on public dataset 1–3. The values of R², MAE, RMSE refers to the mean and standard deviation across the folds

Dataset	Methods	R²	MAE	RMSE
Dataset 1	GraphRXN-concat	0.951 ± 0.004	4.3 ± 0.1	6.0 ± 0.2
	GraphRXN-sum	0.938 ± 0.006	4.9 ± 0.2	6.8 ± 0.3
	Yield-BERT	0.951 ± 0.005	4.0 ± 0.2	6.0 ± 0.3
	DeepReac +	0.922 ± 0.019	5.2 ± 0.6	7.5 ± 0.9
Dataset 2	GraphRXN-concat	0.844 ± 0.007	7.9 ± 0.1	11.1 ± 0.3
	GraphRXN-sum	0.838 ± 0.009	8.1 ± 0.2	11.3 ± 0.4
	Yield-BERT	0.815 ± 0.013	8.1 ± 0.4	12.1 ± 0.5
	DeepReac +	0.827 ± 0.017	8.1 ± 0.4	11.7 ± 0.6
Dataset 3	GraphRXN-concat	0.892 ± 0.008	0.16 ± 0.01	0.23 ± 0.01
	GraphRXN-sum	0.881 ± 0.013	0.18 ± 0.01	0.24 ± 0.02
	Yield-BERT	0.886 ± 0.010	0.16 ± 0.01	0.24 ± 0.01
	DeepReac +	0.853 ± 0.024	0.18 ± 0.01	0.25 ± 0.02

ISSN: 1758-2946