In the attached example there is the constant feature collision_energy_aligned which gets assigned a large weight.
Likely due to a floating point precision error when calculating the standard deviation which is 4.3e-15 instead of 1 which should be the default for constant features.
Normalization factors
Avg 2.5 1.6e+03 6.3 16 0.39
Stdv 0.77 3.9e+02 0.49 22 4.3e-15
Learned normalized SVM weights for the 3 cross-validation splits:
Split1 Split2 Split3 FeatureName
1.3036 1.1007 -0.0034 KR
0.2208 -0.1305 0.0015 Mass
1.0966 -0.0478 -0.0033 RT
1.5000 0.3854 -0.0026 abs_rt_diff
-2.6710 -1.4817 -0.4983 collision_energy_aligned
-2.6710 -1.4817 -0.4983 m0