learning_rate: 0.00015
train_batch_size: 64
eval_batch_size: 1024
seed: 2
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
num_epochs: 3
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Accuracy	Map Mae	Bayes Mae	Expected Score Mae
1.3607	0.1411	500	1.2658	0.5914	0.4723	0.4964	0.6203
1.2648	0.2822	1000	1.2491	0.6240	0.4297	0.4630	0.5999
1.2515	0.4233	1500	1.2393	0.6329	0.4232	0.4383	0.5722
1.2489	0.5643	2000	1.2392	0.6356	0.4233	0.4168	0.5695
1.2429	0.7054	2500	1.2324	0.6436	0.4075	0.4127	0.5582
1.2369	0.8465	3000	1.2311	0.6448	0.4047	0.4107	0.5588
1.2342	0.9876	3500	1.2310	0.6415	0.4103	0.4070	0.5557
1.2260	1.1287	4000	1.2275	0.6456	0.3999	0.4053	0.5606
1.2269	1.2698	4500	1.2272	0.6475	0.3980	0.4077	0.5572
1.2285	1.4108	5000	1.2287	0.6488	0.3976	0.3979	0.5377
1.2251	1.5519	5500	1.2249	0.6497	0.3984	0.3971	0.5404
1.2227	1.6930	6000	1.2249	0.6491	0.4021	0.3927	0.5444
1.2222	1.8341	6500	1.2251	0.6502	0.3973	0.3929	0.5375
1.2211	1.9752	7000	1.2226	0.6496	0.3903	0.4033	0.5522
1.2161	2.1163	7500	1.2237	0.6532	0.3942	0.3946	0.5418
1.2142	2.2573	8000	1.2239	0.6553	0.3908	0.3905	0.5358
1.2139	2.3984	8500	1.2215	0.6547	0.3900	0.3937	0.5402
1.2140	2.5395	9000	1.2212	0.6544	0.3904	0.3916	0.5404
1.2118	2.6806	9500	1.2221	0.6548	0.3912	0.3932	0.5387
1.2118	2.8217	10000	1.2215	0.6549	0.3902	0.3931	0.5390
1.2146	2.9628	10500	1.2215	0.6550	0.3898	0.3933	0.5398
1.2146	3.0	10632	1.2215	0.6549	0.3899	0.3933	0.5397

MichaHenh/cil-ordinal-soft-label-ce-seed2

cil-ordinal-soft-label-ce-seed2