SignBart-KArSL03-190

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1930
• Accuracy: 0.9954
• Precision: 0.9960
• Recall: 0.9954

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: 0.0002
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.4
• num_epochs: 200

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall
5.334	1.0	32	5.2921	0.0072	0.0032	0.0072
5.3151	2.0	64	5.2800	0.0079	0.0011	0.0079
5.3118	3.0	96	5.2602	0.0086	0.0008	0.0086
5.2942	4.0	128	5.2328	0.0099	0.0010	0.0099
5.261	5.0	160	5.1984	0.0184	0.0053	0.0184
5.2327	6.0	192	5.1560	0.0224	0.0039	0.0224
5.1974	7.0	224	5.1033	0.0336	0.0114	0.0336
5.1513	8.0	256	5.0404	0.0428	0.0183	0.0428
5.0868	9.0	288	4.9599	0.0664	0.0520	0.0664
5.0105	10.0	320	4.8618	0.1	0.0749	0.1
4.9454	11.0	352	4.7540	0.1329	0.0864	0.1329
4.8377	12.0	384	4.6324	0.1737	0.1195	0.1737
4.739	13.0	416	4.5026	0.2151	0.1798	0.2151
4.6583	14.0	448	4.3775	0.2612	0.2260	0.2612
4.5649	15.0	480	4.2508	0.2961	0.2448	0.2961
4.4109	16.0	512	4.1107	0.3322	0.2717	0.3322
4.347	17.0	544	3.9741	0.3632	0.3231	0.3632
4.1938	18.0	576	3.8278	0.3921	0.3451	0.3921
4.0716	19.0	608	3.6688	0.4263	0.3863	0.4263
3.923	20.0	640	3.5112	0.4664	0.4293	0.4664
3.8664	21.0	672	3.3628	0.5072	0.4882	0.5072
3.6328	22.0	704	3.1895	0.5651	0.5719	0.5651
3.6225	23.0	736	3.0305	0.6007	0.6039	0.6007
3.4433	24.0	768	2.8734	0.6428	0.6566	0.6428
3.225	25.0	800	2.6833	0.6849	0.7044	0.6849
3.078	26.0	832	2.4971	0.7164	0.7337	0.7164
3.0282	27.0	864	2.3312	0.7592	0.7881	0.7592
2.8731	28.0	896	2.1699	0.7882	0.8204	0.7882
2.7194	29.0	928	1.9987	0.8164	0.8572	0.8164
2.5895	30.0	960	1.8444	0.8428	0.8806	0.8428
2.469	31.0	992	1.6913	0.8664	0.9034	0.8664
2.2569	32.0	1024	1.5361	0.8849	0.9175	0.8849
2.1404	33.0	1056	1.3969	0.9059	0.9307	0.9059
2.1261	34.0	1088	1.2720	0.9145	0.9375	0.9145
2.022	35.0	1120	1.1560	0.9414	0.9532	0.9414
1.8954	36.0	1152	1.0580	0.9395	0.9532	0.9395
1.7565	37.0	1184	0.9626	0.9592	0.9681	0.9592
1.7046	38.0	1216	0.8776	0.9566	0.9681	0.9566
1.6673	39.0	1248	0.7934	0.9664	0.9727	0.9664
1.6228	40.0	1280	0.7310	0.9717	0.9770	0.9717
1.4788	41.0	1312	0.6673	0.9671	0.9737	0.9671
1.3849	42.0	1344	0.6016	0.9776	0.9817	0.9776
1.4186	43.0	1376	0.5510	0.9796	0.9831	0.9796
1.431	44.0	1408	0.5103	0.9757	0.9784	0.9757
1.1953	45.0	1440	0.4606	0.9816	0.9844	0.9816
1.2252	46.0	1472	0.4225	0.9849	0.9872	0.9849
1.1936	47.0	1504	0.3792	0.9875	0.9902	0.9875
1.103	48.0	1536	0.3544	0.9842	0.9874	0.9842
0.9981	49.0	1568	0.3273	0.9882	0.9898	0.9882
0.896	50.0	1600	0.2914	0.9882	0.99	0.9882
0.8749	51.0	1632	0.2644	0.9934	0.9943	0.9934
0.9732	52.0	1664	0.2459	0.9908	0.9922	0.9908
0.9124	53.0	1696	0.2243	0.9914	0.9931	0.9914
0.8274	54.0	1728	0.2063	0.9934	0.9944	0.9934
0.7723	55.0	1760	0.1930	0.9954	0.9960	0.9954

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.19.1