Complete result settings for 20 Newsgroup dataset (table columns are sortable)
The experiments are described in Rooshenas and Lowd, Discriminative Structure Learning of Arithmetic Circuits, AIStats 16

Standard Deviation L1 Penalty Train CLL Validation CLL Test CLL Node# Edge# Feature# Learning Time
0.1 2 -26.224995 -23.780677 -29.442480 92615 180971 9358 53038.732886
0.5 0.5 -26.214539 -23.783153 -29.458774 92615 180971 9358 56076.109135
0.1 0.5 -26.214663 -23.782529 -29.459059 92615 180971 9358 57000.377624
0.5 2 -26.225267 -23.779845 -29.442740 92615 180971 9358 56864.978209
0.1 1 -26.218666 -23.781728 -29.455816 92615 180971 9358 58035.753223
0.5 1 -26.218567 -23.782212 -29.455083 92615 180971 9358 65288.251676
0.5 0.1 -26.229132 -23.793331 -29.479745 92615 180971 9358 38867.181289
0.1 0.1 -26.229140 -23.793323 -29.479980 92615 180971 9358 39981.129943