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

Split Penalty	Standard Deviation	L1 Penalty	Train LL	Validation LL	Test LL	Node#	Edge#	Feature#	Learning Time
10	0.5	1	-6.015215	-5.978279	-6.024370	17271	37987	380	55.720528
10	0.1	0.1	-6.017510	-5.971981	-6.028945	17781	39169	414	69.572423
5	0.1	0.1	-5.997362	-5.961181	-6.012205	22502	49514	550	70.788238
5	0.1	1	-5.995495	-5.955038	-6.010212	23061	50781	566	74.032745
10	0.5	0.1	-6.015458	-5.974007	-6.017623	18004	39653	384	76.326396
10	0.1	2	-6.016938	-5.980298	-6.030138	19618	43224	354	83.000381
10	0.5	2	-6.014467	-5.978513	-6.024752	19385	43002	376	91.039159
10	0.1	1	-6.017169	-5.984284	-6.027594	19371	42699	388	90.448249
5	0.5	0.1	-5.990309	-5.954208	-6.019967	22801	50028	524	94.525629
5	0.5	1	-5.992335	-5.948668	-6.007810	22324	49366	546	104.159165
5	0.1	2	-6.000011	-5.959797	-6.014303	24566	53343	490	106.330834
5	0.5	2	-5.995405	-5.965584	-6.014937	25642	56638	510	123.933158
2	0.1	1	-5.980254	-5.942791	-6.003117	28660	62279	772	115.648418
2	0.5	0.1	-5.972368	-5.946560	-6.001371	29255	63612	812	146.558719
2	0.1	2	-5.981908	-5.951267	-6.003761	36668	79592	852	169.947163
2	0.1	0.1	-5.975313	-5.942313	-6.005768	31071	68075	826	156.950139
2	0.5	2	-5.980321	-5.944992	-5.997035	25744	55812	724	132.085919
2	0.5	1	-5.974816	-5.944132	-6.001134	28246	62155	808	130.095221
1	0.5	2	-5.967330	-5.949079	-6.003139	38889	84418	1010	245.398693
1	0.5	1	-5.960654	-5.947322	-6.001921	47622	103816	1134	309.062015
1	0.5	0.1	-5.953199	-5.956266	-6.008703	51532	112135	1232	315.378054
1	0.1	1	-5.967355	-5.948409	-6.000732	36812	79816	1018	124.739036
1	0.1	2	-5.971131	-5.949984	-5.999679	38620	83768	992	251.493766
1	0.1	0.1	-5.961153	-5.939971	-6.003436	44581	97531	1170	295.428087