uspto_mixed dataset

[feiyang-cai]

Hi,

Thanks for your nice work!

I noticed that some of the reactions with multi-products are simplified to single-product reaction, which is not consistent with the original mit dataset.

For example, in one of the reactions in the test,
original products CC(c1cccc(N)c1)S(=O)(=O)[O-].CCCC[N+](CCCC)(CCCC)CCCC are simplified to CC(c1cccc(N)c1)S(=O)(=O)[O-].

Actually, there is in total of 903 reactions simplified.

I am wondering if there is any reason to simplify it.

Thanks!