Hi David,

here are some preliminary results on CIFAR10. As discussed, there are some experiments with different seed_optimization and seed_weights.

Is this kind of what you had in mind?

PS: I didn't fit a gaussian to the scatter plots yet, I just did a linear regression.

Results: fixed seed_weight, changing seed_optimization

	so	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99
val	sw=0	79.33	79.22	79.40	79.18	79.44	79.45	79.65	79.29	79.37	79.16	79.23	79.54	79.23	79.13	79.14	79.24	79.22	79.68	79.21	78.98	78.94	79.09	79.39	79.21	79.65	79.67	79.29	79.38	79.25	79.76	79.10	79.12	79.07	79.08	79.55	79.33	79.43	79.68	79.44	79.19	79.14	79.24	79.66	79.17	79.17	79.14	79.46	79.25	79.58	79.02	79.32	78.95	79.22	79.68	79.16	79.45	79.94	79.31	78.99	79.64	79.50	79.38	79.21	79.74	79.81	79.10	79.16	79.87	79.25	79.13	78.97	79.30	79.57	79.49	79.80	79.35	79.39	79.00	79.57	79.31	79.52	79.22	79.46	79.20	79.23	79.47	79.30	79.14	79.49	79.35	79.40	79.74	79.31	79.52	79.31	79.46	79.23	79.62	79.26	79.43
test	sw=0	77.99	78.41	78.07	77.93	78.66	78.71	78.75	78.38	78.27	78.68	78.69	78.35	78.06	78.24	78.16	78.32	78.14	78.55	78.64	78.41	78.25	78.02	77.71	78.36	78.08	78.62	78.56	78.16	78.34	78.52	78.28	78.23	78.56	78.28	78.67	78.47	78.68	78.41	79.17	79.01	78.39	78.85	78.18	78.79	78.06	78.89	78.20	77.90	79.29	77.54	78.32	78.10	78.83	79.24	77.95	78.89	78.50	78.34	77.63	79.19	78.92	78.52	78.92	78.25	78.23	78.29	78.28	79.17	78.87	78.25	77.99	78.92	78.58	77.51	78.37	79.48	78.57	77.60	78.15	78.56	78.08	78.16	78.27	78.17	78.25	78.39	78.10	78.63	78.95	78.96	78.99	79.07	78.40	78.98	78.18	78.96	78.50	78.72	79.01	78.30

Results: fixed seed_optimization, changing seed_weight

	sw	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99
val	so=0	79.33	79.56	78.95	79.29	79.54	79.15	79.30	79.39	78.99	79.30	79.48	79.26	79.41	79.05	79.54	79.35	79.48	79.38	79.29	79.05	79.24	79.40	79.02	79.47	79.56	79.53	79.47	79.08	79.41	79.82	79.50	79.59	79.68	79.16	79.15	79.34	79.74	79.90	79.45	79.37	79.51	79.29	79.62	79.42	79.18	79.26	79.43	79.21	79.55	79.45	79.71	79.32	79.20	79.54	79.31	79.42	79.06	79.64	79.43	79.50	79.48	79.46	79.25	79.49	79.77	79.11	79.37	79.83	79.61	79.34	79.38	79.53	79.66	79.27	79.42	79.67	79.65	79.78	79.62	79.59	79.39	79.51	79.69	78.95	79.53	79.38	79.43	79.55	79.35	79.57	79.41	79.62	79.13	79.91	79.56	79.41	79.05	79.19	79.15	79.49
test	so=0	77.99	77.68	77.66	78.66	78.23	78.44	79.23	78.55	77.77	78.38	78.38	78.04	78.27	77.87	78.58	78.40	78.40	78.83	78.60	77.69	77.68	78.66	78.32	78.62	78.58	77.79	78.54	78.70	78.51	78.38	78.39	78.52	78.07	78.09	77.67	79.08	78.23	78.10	78.41	78.53	78.24	78.88	78.49	78.96	77.86	77.70	78.67	78.21	78.93	78.42	78.23	78.39	78.24	78.89	78.36	78.15	78.71	78.68	78.39	78.86	78.85	78.89	78.73	78.47	78.73	78.27	78.71	79.06	78.18	78.20	78.73	78.10	78.79	78.42	78.57	78.49	78.68	78.30	78.81	78.46	78.07	78.60	79.01	78.29	78.82	78.81	78.45	78.31	78.81	78.80	78.52	78.38	78.46	78.35	78.82	78.28	78.12	78.22	78.31	78.43