vit-base-patch16-224-in21k-finetune-os300_norm

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3499
• Accuracy: 0.8577

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: 0.005
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.038	0.98	11	0.7215	0.6568
0.7212	1.96	22	0.7280	0.6568
0.7201	2.93	33	0.7285	0.6568
0.7308	4.0	45	0.7297	0.6568
0.7341	4.98	56	0.7277	0.6568
0.7255	5.96	67	0.7350	0.6568
0.7274	6.93	78	0.7258	0.6568
0.7189	8.0	90	0.7205	0.6568
0.7194	8.98	101	0.7117	0.6568
0.7437	9.96	112	0.7340	0.6568
0.7578	10.93	123	0.7317	0.6568
0.7307	12.0	135	0.7288	0.6568
0.7279	12.98	146	0.7246	0.6568
0.727	13.96	157	0.7166	0.6568
0.7161	14.93	168	0.7306	0.5117
0.6775	16.0	180	0.6360	0.6568
0.6487	16.98	191	0.6166	0.7113
0.607	17.96	202	0.5871	0.7240
0.5961	18.93	213	0.5606	0.7183
0.5681	20.0	225	0.5459	0.7381
0.5756	20.98	236	0.5375	0.7481
0.5666	21.96	247	0.5720	0.7042
0.5658	22.93	258	0.5127	0.7481
0.5461	24.0	270	0.5254	0.7360
0.5484	24.98	281	0.5124	0.7431
0.5442	25.96	292	0.5665	0.7282
0.5573	26.93	303	0.5019	0.7594
0.535	28.0	315	0.5112	0.7792
0.5319	28.98	326	0.4729	0.7856
0.4953	29.96	337	0.6292	0.7318
0.5408	30.93	348	0.5083	0.7877
0.5215	32.0	360	0.5131	0.7799
0.5291	32.98	371	0.4867	0.7983
0.4971	33.96	382	0.4742	0.7962
0.5004	34.93	393	0.4930	0.7806
0.4868	36.0	405	0.4550	0.8061
0.4784	36.98	416	0.4667	0.7912
0.469	37.96	427	0.4915	0.7856
0.455	38.93	438	0.5016	0.7537
0.4903	40.0	450	0.4874	0.7877
0.4904	40.98	461	0.5222	0.7629
0.513	41.96	472	0.4772	0.7877
0.4913	42.93	483	0.5386	0.7629
0.5216	44.0	495	0.4830	0.7827
0.4931	44.98	506	0.4692	0.7948
0.4835	45.96	517	0.4941	0.7757
0.5035	46.93	528	0.4716	0.7884
0.5068	48.0	540	0.5210	0.7841
0.5207	48.98	551	0.4656	0.8132
0.4753	49.96	562	0.4529	0.8025
0.4718	50.93	573	0.4403	0.8075
0.4757	52.0	585	0.4305	0.8132
0.4352	52.98	596	0.4104	0.8245
0.4349	53.96	607	0.4390	0.8125
0.4508	54.93	618	0.4409	0.8011
0.4596	56.0	630	0.4131	0.8323
0.4321	56.98	641	0.4257	0.8188
0.4433	57.96	652	0.4421	0.7969
0.4423	58.93	663	0.4430	0.7990
0.446	60.0	675	0.4328	0.8181
0.425	60.98	686	0.4385	0.8011
0.4363	61.96	697	0.4225	0.8139
0.4358	62.93	708	0.4114	0.8224
0.415	64.0	720	0.4110	0.8174
0.423	64.98	731	0.4090	0.8238
0.4161	65.96	742	0.4011	0.8160
0.4103	66.93	753	0.4207	0.8188
0.4254	68.0	765	0.4503	0.8004
0.429	68.98	776	0.4392	0.8033
0.4341	69.96	787	0.4159	0.8209
0.4574	70.93	798	0.4165	0.8224
0.4136	72.0	810	0.3954	0.8337
0.4226	72.98	821	0.3996	0.8301
0.4124	73.96	832	0.4205	0.8089
0.4209	74.93	843	0.4288	0.8146
0.4493	76.0	855	0.4193	0.8167
0.4302	76.98	866	0.4239	0.8132
0.4385	77.96	877	0.4187	0.8160
0.4388	78.93	888	0.4379	0.8047
0.4294	80.0	900	0.4048	0.8309
0.4207	80.98	911	0.4287	0.8139
0.4316	81.96	922	0.4183	0.8202
0.4283	82.93	933	0.4091	0.8224
0.4227	84.0	945	0.4070	0.8231
0.4335	84.98	956	0.4184	0.8224
0.4433	85.96	967	0.4148	0.8132
0.4287	86.93	978	0.4188	0.8167
0.4327	88.0	990	0.4091	0.8224
0.427	88.98	1001	0.4118	0.8202
0.4194	89.96	1012	0.4220	0.8153
0.4213	90.93	1023	0.4195	0.8096
0.4288	92.0	1035	0.4023	0.8188
0.4123	92.98	1046	0.4005	0.8393
0.4172	93.96	1057	0.3812	0.8309
0.4109	94.93	1068	0.3838	0.8294
0.4128	96.0	1080	0.3878	0.8294
0.3976	96.98	1091	0.4023	0.8259
0.4097	97.96	1102	0.3979	0.8153
0.4059	98.93	1113	0.3953	0.8294
0.4011	100.0	1125	0.3804	0.8344
0.4126	100.98	1136	0.3915	0.8259
0.425	101.96	1147	0.4140	0.8160
0.4066	102.93	1158	0.4207	0.8238
0.4265	104.0	1170	0.4016	0.8259
0.4225	104.98	1181	0.4059	0.8252
0.4201	105.96	1192	0.3980	0.8309
0.408	106.93	1203	0.4171	0.8202
0.422	108.0	1215	0.4475	0.8096
0.4251	108.98	1226	0.4139	0.8224
0.4261	109.96	1237	0.4113	0.8167
0.4147	110.93	1248	0.4355	0.8089
0.4407	112.0	1260	0.4453	0.8146
0.4167	112.98	1271	0.3987	0.8372
0.4152	113.96	1282	0.4008	0.8273
0.3952	114.93	1293	0.3843	0.8351
0.4159	116.0	1305	0.3949	0.8330
0.4014	116.98	1316	0.4113	0.8040
0.4203	117.96	1327	0.3988	0.8309
0.4159	118.93	1338	0.4037	0.8351
0.4065	120.0	1350	0.3847	0.8393
0.3938	120.98	1361	0.4023	0.8280
0.4202	121.96	1372	0.4015	0.8301
0.4316	122.93	1383	0.4156	0.8174
0.416	124.0	1395	0.3924	0.8344
0.4141	124.98	1406	0.3839	0.8358
0.4157	125.96	1417	0.3940	0.8224
0.3906	126.93	1428	0.3826	0.8287
0.4051	128.0	1440	0.3807	0.8316
0.3835	128.98	1451	0.3866	0.8386
0.3976	129.96	1462	0.3832	0.8457
0.3939	130.93	1473	0.3745	0.8351
0.3862	132.0	1485	0.3897	0.8408
0.3919	132.98	1496	0.3841	0.8429
0.3928	133.96	1507	0.3744	0.8507
0.3976	134.93	1518	0.3610	0.8535
0.3834	136.0	1530	0.3711	0.8422
0.3827	136.98	1541	0.3860	0.8422
0.4036	137.96	1552	0.3973	0.8301
0.3862	138.93	1563	0.3720	0.8429
0.3876	140.0	1575	0.3701	0.8478
0.3941	140.98	1586	0.3579	0.8500
0.3692	141.96	1597	0.3609	0.8521
0.3791	142.93	1608	0.3666	0.8493
0.3774	144.0	1620	0.3601	0.8521
0.3708	144.98	1631	0.3592	0.8549
0.3943	145.96	1642	0.3593	0.8493
0.3856	146.93	1653	0.3686	0.8429
0.381	148.0	1665	0.3755	0.8429
0.3965	148.98	1676	0.3698	0.8471
0.3862	149.96	1687	0.3641	0.8485
0.3825	150.93	1698	0.3652	0.8528
0.3751	152.0	1710	0.3672	0.8422
0.3812	152.98	1721	0.3626	0.8507
0.3805	153.96	1732	0.3615	0.8493
0.3755	154.93	1743	0.3678	0.8500
0.3802	156.0	1755	0.3682	0.8478
0.3781	156.98	1766	0.3802	0.8485
0.3845	157.96	1777	0.3753	0.8507
0.3893	158.93	1788	0.3694	0.8485
0.3676	160.0	1800	0.3652	0.8493
0.4114	160.98	1811	0.4020	0.8309
0.39	161.96	1822	0.3615	0.8528
0.3831	162.93	1833	0.3570	0.8535
0.3651	164.0	1845	0.3642	0.8401
0.3662	164.98	1856	0.3557	0.8577
0.3878	165.96	1867	0.3650	0.8457
0.376	166.93	1878	0.3601	0.8500
0.3724	168.0	1890	0.3617	0.8570
0.3661	168.98	1901	0.3677	0.8535
0.3869	169.96	1912	0.3617	0.8500
0.3717	170.93	1923	0.3594	0.8436
0.3698	172.0	1935	0.3632	0.8514
0.3761	172.98	1946	0.3614	0.8471
0.3847	173.96	1957	0.3566	0.8535
0.3716	174.93	1968	0.3570	0.8528
0.3695	176.0	1980	0.3557	0.8556
0.3702	176.98	1991	0.3544	0.8556
0.372	177.96	2002	0.3522	0.8542
0.3648	178.93	2013	0.3562	0.8493
0.3744	180.0	2025	0.3577	0.8507
0.3546	180.98	2036	0.3524	0.8535
0.3613	181.96	2047	0.3478	0.8528
0.3581	182.93	2058	0.3534	0.8549
0.3709	184.0	2070	0.3637	0.8521
0.3699	184.98	2081	0.3544	0.8549
0.3701	185.96	2092	0.3506	0.8613
0.3634	186.93	2103	0.3559	0.8592
0.3668	188.0	2115	0.3510	0.8585
0.3629	188.98	2126	0.3485	0.8592
0.3544	189.96	2137	0.3478	0.8627
0.3714	190.93	2148	0.3512	0.8592
0.3681	192.0	2160	0.3522	0.8592
0.3466	192.98	2171	0.3523	0.8570
0.3727	193.96	2182	0.3504	0.8606
0.3564	194.93	2193	0.3501	0.8577
0.3616	195.56	2200	0.3499	0.8577

Framework versions

• Transformers 4.39.0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2