test-hasy-5

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

• Loss: 0.6861
• Accuracy: 0.8067

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: 8
• eval_batch_size: 8
• seed: 1787
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.9645	1.0	541	3.4295	0.3971
3.4258	2.0	1082	2.8790	0.4782
3.04	3.0	1623	2.4893	0.5468
2.793	4.0	2164	2.2006	0.5738
2.5551	5.0	2705	1.9056	0.6341
2.3662	6.0	3246	1.7023	0.6632
2.1965	7.0	3787	1.5740	0.6798
2.1397	8.0	4328	1.4561	0.6944
1.9955	9.0	4869	1.3203	0.7235
1.9282	10.0	5410	1.2246	0.7380
1.8368	11.0	5951	1.1823	0.7380
1.812	12.0	6492	1.1298	0.7214
1.7195	13.0	7033	1.0423	0.7484
1.6314	14.0	7574	1.0077	0.7422
1.5979	15.0	8115	1.0051	0.7464
1.5656	16.0	8656	0.9325	0.7692
1.5414	17.0	9197	0.8889	0.7734
1.5342	18.0	9738	0.9073	0.7484
1.4898	19.0	10279	0.8426	0.7713
1.4731	20.0	10820	0.8625	0.7443
1.451	21.0	11361	0.8015	0.7630
1.4578	22.0	11902	0.8520	0.7588
1.4126	23.0	12443	0.7928	0.7713
1.3626	24.0	12984	0.7544	0.7838
1.3694	25.0	13525	0.7699	0.7775
1.3612	26.0	14066	0.7602	0.7775
1.2963	27.0	14607	0.7532	0.7713
1.3009	28.0	15148	0.7013	0.7921
1.2598	29.0	15689	0.7085	0.7796
1.2565	30.0	16230	0.7023	0.7775
1.2735	31.0	16771	0.7048	0.7775
1.2743	32.0	17312	0.6794	0.7921
1.2441	33.0	17853	0.6932	0.7859
1.2282	34.0	18394	0.7039	0.7942
1.2204	35.0	18935	0.6861	0.8067
1.1808	36.0	19476	0.6590	0.7963
1.1928	37.0	20017	0.6784	0.7817
1.1914	38.0	20558	0.6559	0.7963
1.1856	39.0	21099	0.6769	0.7963
1.1585	40.0	21640	0.6498	0.8004
1.1713	41.0	22181	0.6447	0.7921
1.1183	42.0	22722	0.6748	0.7713
1.1564	43.0	23263	0.6545	0.7921
1.1215	44.0	23804	0.6690	0.7879
1.1008	45.0	24345	0.6598	0.7879
1.1344	46.0	24886	0.6550	0.8025
1.126	47.0	25427	0.6521	0.7859
1.125	48.0	25968	0.6813	0.7817
1.0855	49.0	26509	0.6419	0.7859
1.0452	50.0	27050	0.6551	0.8004
1.0626	51.0	27591	0.6675	0.7921
1.0155	52.0	28132	0.6946	0.7921
1.0319	53.0	28673	0.6942	0.7796
1.0488	54.0	29214	0.6496	0.7983
1.0558	55.0	29755	0.6465	0.8046
0.9913	56.0	30296	0.6654	0.7921
1.0555	57.0	30837	0.6561	0.7963
0.9803	58.0	31378	0.6732	0.7942
1.0393	59.0	31919	0.6893	0.7817
0.9677	60.0	32460	0.6824	0.8046
1.0082	61.0	33001	0.6618	0.7942
1.0096	62.0	33542	0.6691	0.7838
0.9685	63.0	34083	0.6793	0.8025
0.9847	64.0	34624	0.6895	0.7838
0.9639	65.0	35165	0.7297	0.7734
0.9776	66.0	35706	0.6561	0.7921
1.0074	67.0	36247	0.6999	0.7775
0.9466	68.0	36788	0.6881	0.7942
0.9425	69.0	37329	0.6806	0.7963
0.9594	70.0	37870	0.7202	0.7900
0.9311	71.0	38411	0.7162	0.7755
0.9429	72.0	38952	0.7284	0.7921
0.9666	73.0	39493	0.6871	0.7963
0.945	74.0	40034	0.6779	0.7942
0.9387	75.0	40575	0.7358	0.7942
0.9132	76.0	41116	0.7044	0.7942
0.9181	77.0	41657	0.7041	0.7963
0.9218	78.0	42198	0.6986	0.7942
0.8621	79.0	42739	0.6909	0.8004
0.9236	80.0	43280	0.7136	0.7983
0.8667	81.0	43821	0.7009	0.8025
0.8856	82.0	44362	0.7128	0.7921
0.917	83.0	44903	0.7135	0.7983
0.8835	84.0	45444	0.7295	0.7900
0.8879	85.0	45985	0.7450	0.7900
0.8764	86.0	46526	0.7362	0.7942
0.8674	87.0	47067	0.7232	0.7942
0.8583	88.0	47608	0.7408	0.7942
0.881	89.0	48149	0.7378	0.8004
0.8668	90.0	48690	0.7473	0.7900
0.8779	91.0	49231	0.7438	0.7983
0.8717	92.0	49772	0.7390	0.8004
0.8781	93.0	50313	0.7474	0.7983
0.8845	94.0	50854	0.7446	0.7900
0.8623	95.0	51395	0.7316	0.7921
0.8341	96.0	51936	0.7457	0.7879
0.8766	97.0	52477	0.7436	0.7921
0.8681	98.0	53018	0.7484	0.7900
0.8635	99.0	53559	0.7392	0.7942
0.8091	100.0	54100	0.7391	0.7921

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2