vit-base_rvl-cdip-small_rvl_cdip-NK1000_kd_MSE_rand

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the None dataset. It achieves the following results on the evaluation set:

Loss: 1.8136
Accuracy: 0.6262
Brier Loss: 0.5539
Nll: 2.6914
F1 Micro: 0.6262
F1 Macro: 0.6298
Ece: 0.1916
Aurc: 0.1624

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
4.1455	1.0	1000	3.9403	0.1898	0.8859	5.5812	0.1898	0.1327	0.0656	0.6790
3.4397	2.0	2000	3.3498	0.346	0.8036	4.5993	0.346	0.2909	0.1060	0.4698
2.983	3.0	3000	3.1592	0.384	0.8070	3.7537	0.384	0.3451	0.1795	0.4280
2.7406	4.0	4000	2.7614	0.4447	0.6956	3.6359	0.4447	0.4361	0.0840	0.3307
2.5937	5.0	5000	2.7073	0.4462	0.7000	3.5795	0.4462	0.4321	0.1292	0.3212
2.3878	6.0	6000	2.3935	0.5012	0.6651	3.1888	0.5012	0.4842	0.1429	0.2829
2.2284	7.0	7000	2.3189	0.5022	0.6378	3.0628	0.5022	0.5109	0.1027	0.2630
2.0759	8.0	8000	2.3408	0.4993	0.6550	3.3921	0.4993	0.4923	0.1398	0.2640
1.9764	9.0	9000	2.0531	0.5563	0.5946	2.9188	0.5563	0.5619	0.1202	0.2170
1.8232	10.0	10000	2.1083	0.5505	0.6295	3.0945	0.5505	0.5445	0.1794	0.2322
1.7049	11.0	11000	2.0447	0.5653	0.6142	2.9718	0.5653	0.5605	0.1684	0.2207
1.6182	12.0	12000	2.0684	0.5637	0.6462	3.0100	0.5637	0.5595	0.2095	0.2272
1.4886	13.0	13000	1.9374	0.5735	0.6132	2.9415	0.5735	0.5806	0.1874	0.2042
1.3538	14.0	14000	2.0147	0.5895	0.6174	3.0835	0.5895	0.5851	0.1966	0.2109
1.2304	15.0	15000	1.9766	0.5867	0.6203	3.1471	0.5867	0.5846	0.2229	0.2091
1.1124	16.0	16000	1.8998	0.6008	0.6044	2.9169	0.6008	0.5943	0.2144	0.1911
1.0197	17.0	17000	1.9309	0.5955	0.6123	3.1166	0.5955	0.5979	0.2299	0.1876
0.8763	18.0	18000	1.9741	0.5952	0.6316	3.2227	0.5952	0.5971	0.2439	0.1957
0.8042	19.0	19000	1.9944	0.592	0.6318	3.1537	0.592	0.5898	0.2439	0.2024
0.7059	20.0	20000	1.9259	0.6082	0.6124	3.0665	0.6082	0.6093	0.2344	0.1889
0.632	21.0	21000	1.9444	0.6095	0.6148	3.0133	0.6095	0.6111	0.2281	0.1917
0.5641	22.0	22000	1.9830	0.5968	0.6282	3.0999	0.5968	0.5984	0.2442	0.1913
0.5138	23.0	23000	2.0190	0.5962	0.6331	3.1937	0.5962	0.5966	0.2501	0.2033
0.457	24.0	24000	1.9340	0.6075	0.6151	2.9559	0.6075	0.6096	0.2333	0.1888
0.3999	25.0	25000	1.9742	0.6048	0.6285	3.0455	0.6048	0.6080	0.2461	0.1939
0.3629	26.0	26000	1.9308	0.6142	0.6027	3.1686	0.6142	0.6169	0.2244	0.1850
0.3132	27.0	27000	1.9468	0.6175	0.6076	3.0271	0.6175	0.6189	0.2374	0.1863
0.2818	28.0	28000	1.9392	0.6095	0.6152	3.0499	0.6095	0.6079	0.2422	0.1894
0.2584	29.0	29000	1.8976	0.6202	0.6040	2.9355	0.6202	0.6204	0.2340	0.1834
0.228	30.0	30000	1.9111	0.617	0.6020	3.0272	0.617	0.6192	0.2336	0.1780
0.2041	31.0	31000	1.8513	0.6272	0.5835	2.8808	0.6272	0.6293	0.2222	0.1733
0.1834	32.0	32000	1.8501	0.6262	0.5782	2.8280	0.6262	0.6275	0.2142	0.1702
0.1613	33.0	33000	1.8250	0.6292	0.5712	2.8863	0.6292	0.6338	0.2021	0.1691
0.1437	34.0	34000	1.8457	0.6228	0.5773	2.9046	0.6228	0.6232	0.2114	0.1717
0.1275	35.0	35000	1.8088	0.6315	0.5646	2.8124	0.6315	0.6328	0.2039	0.1638
0.1127	36.0	36000	1.8204	0.6335	0.5647	2.7943	0.6335	0.6373	0.1993	0.1661
0.1026	37.0	37000	1.8070	0.631	0.5641	2.7537	0.631	0.6326	0.2015	0.1634
0.0894	38.0	38000	1.8068	0.63	0.5606	2.7461	0.63	0.6317	0.1998	0.1634
0.0785	39.0	39000	1.7894	0.6312	0.5550	2.7333	0.6312	0.6351	0.1963	0.1599
0.0696	40.0	40000	1.7996	0.6288	0.5607	2.7489	0.6288	0.6334	0.1986	0.1645
0.0626	41.0	41000	1.7963	0.6328	0.5532	2.7232	0.6328	0.6349	0.1933	0.1632
0.055	42.0	42000	1.7959	0.6268	0.5556	2.6877	0.6268	0.6298	0.1957	0.1617
0.0475	43.0	43000	1.8018	0.632	0.5522	2.7232	0.632	0.6354	0.1934	0.1598
0.0419	44.0	44000	1.7930	0.6325	0.5507	2.6842	0.6325	0.6361	0.1906	0.1612
0.0367	45.0	45000	1.8064	0.6265	0.5577	2.6772	0.6265	0.6299	0.1994	0.1632
0.0328	46.0	46000	1.8044	0.6228	0.5524	2.6611	0.6228	0.6263	0.1971	0.1620
0.0289	47.0	47000	1.8101	0.6248	0.5544	2.6841	0.6248	0.6284	0.1943	0.1624
0.0265	48.0	48000	1.8088	0.6242	0.5531	2.6870	0.6242	0.6283	0.1943	0.1622
0.0238	49.0	49000	1.8107	0.6255	0.5533	2.7007	0.6255	0.6292	0.1923	0.1621
0.022	50.0	50000	1.8136	0.6262	0.5539	2.6914	0.6262	0.6298	0.1916	0.1624

Framework versions

Transformers 4.26.1
Pytorch 1.13.1.post200
Datasets 2.9.0
Tokenizers 0.13.2

jordyvl/vit-base_rvl-cdip-small_rvl_cdip-NK1000_kd_MSE_rand

vit-base_rvl-cdip-small_rvl_cdip-NK1000_kd_MSE_rand

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions