ryan03312024_lr_2e-5_wd_001_v2

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

• Loss: 0.1914
• Ordinal Mae: 0.4198
• Ordinal Accuracy: 0.6843
• Na Accuracy: 0.8505

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Ordinal Mae	Ordinal Accuracy	Na Accuracy
0.4426	0.04	100	0.3707	0.8707	0.3409	0.8076
0.3133	0.07	200	0.3203	0.8545	0.4300	0.7749
0.3349	0.11	300	0.2997	0.8339	0.4593	0.8419
0.3173	0.14	400	0.2870	0.7993	0.4819	0.8660
0.2946	0.18	500	0.2856	0.7690	0.5112	0.8935
0.3002	0.22	600	0.2724	0.7233	0.5345	0.9210
0.2817	0.25	700	0.2657	0.6928	0.5566	0.8625
0.2939	0.29	800	0.2596	0.6425	0.5862	0.7921
0.2525	0.32	900	0.2459	0.6053	0.6047	0.8265
0.2163	0.36	1000	0.2400	0.5777	0.6245	0.8110
0.2181	0.4	1100	0.2339	0.5430	0.6024	0.8763
0.1949	0.43	1200	0.2331	0.5329	0.6286	0.7955
0.214	0.47	1300	0.2424	0.5244	0.6183	0.7629
0.27	0.5	1400	0.2298	0.4995	0.6368	0.7869
0.2117	0.54	1500	0.2301	0.4950	0.6473	0.7784
0.2038	0.58	1600	0.2156	0.4899	0.6550	0.8368
0.1974	0.61	1700	0.2212	0.4639	0.6347	0.8282
0.1916	0.65	1800	0.2151	0.4790	0.6440	0.8797
0.1921	0.69	1900	0.2050	0.4614	0.6609	0.8729
0.1936	0.72	2000	0.2061	0.4566	0.6496	0.8574
0.1939	0.76	2100	0.2294	0.4657	0.6363	0.9089
0.257	0.79	2200	0.2054	0.4567	0.6527	0.8608
0.2236	0.83	2300	0.2044	0.4542	0.6640	0.8763
0.1925	0.87	2400	0.2085	0.4463	0.6887	0.8076
0.1657	0.9	2500	0.2034	0.4392	0.6769	0.8522
0.1723	0.94	2600	0.1957	0.4257	0.6756	0.8385
0.2279	0.97	2700	0.1946	0.4287	0.6740	0.8643
0.1421	1.01	2800	0.1914	0.4198	0.6843	0.8505
0.1116	1.05	2900	0.2019	0.4214	0.6704	0.8230
0.1194	1.08	3000	0.1954	0.4178	0.6807	0.8368
0.1312	1.12	3100	0.1930	0.4166	0.6874	0.8591
0.1836	1.15	3200	0.1989	0.4107	0.6794	0.8643
0.1282	1.19	3300	0.1951	0.4127	0.6971	0.8540
0.1406	1.23	3400	0.1959	0.4036	0.6974	0.8505
0.0929	1.26	3500	0.1969	0.4020	0.6977	0.8454
0.1135	1.3	3600	0.1957	0.4026	0.6982	0.8316
0.1345	1.33	3700	0.1987	0.4107	0.6833	0.8814
0.1198	1.37	3800	0.1969	0.3988	0.6992	0.8522
0.1281	1.41	3900	0.1977	0.4066	0.6966	0.8402
0.1153	1.44	4000	0.2014	0.4091	0.6936	0.8436
0.1485	1.48	4100	0.1965	0.3989	0.7038	0.8385
0.1292	1.51	4200	0.1969	0.3978	0.7031	0.8471
0.1233	1.55	4300	0.1989	0.3993	0.6951	0.8660
0.1128	1.59	4400	0.1998	0.3920	0.6971	0.8522
0.0964	1.62	4500	0.2005	0.3926	0.6982	0.8625
0.1184	1.66	4600	0.2008	0.3860	0.6969	0.8711
0.108	1.69	4700	0.1994	0.3907	0.7020	0.8574
0.129	1.73	4800	0.1985	0.3896	0.7033	0.8591
0.1396	1.77	4900	0.1998	0.3834	0.6984	0.8574
0.1323	1.8	5000	0.1986	0.3844	0.7051	0.8454
0.1079	1.84	5100	0.1974	0.3833	0.7054	0.8402
0.0802	1.88	5200	0.1965	0.3822	0.7074	0.8488
0.1391	1.91	5300	0.1975	0.3809	0.7051	0.8454
0.1183	1.95	5400	0.1973	0.3827	0.7087	0.8351
0.1368	1.98	5500	0.1975	0.3813	0.7082	0.8333

Framework versions

• Transformers 4.39.1
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2