Fig. 3From: MLSolvA: solvation free energy prediction from pairwise atomistic interactions by machine learningTwo-dimensional visualizations on a pre-trained vector from the skip-gram model \(\sum _{\beta } \mathbf {y}_{\beta }\) and b, c extracted molecular feature vector \(\mathbf {v}\) for 15,432 solutes. We reduce the dimensions of each vector using the t-SNE algorithm. The color representation denotes the hydration energy of each pointBack to article page