SignBart-KArSL-ALL-100

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

• Loss: 0.0278
• Accuracy: 0.9942
• Precision: 0.9947
• Recall: 0.9942

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
• num_epochs: 1000

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss	Precision	Recall
5.1417	1.0	50	0.1175	3.9188	0.0961	0.1175
4.323	2.0	100	0.3996	2.7383	0.3779	0.3996
3.5125	3.0	150	0.6454	1.9336	0.6952	0.6454
2.9748	4.0	200	0.7933	1.3981	0.8114	0.7933
2.6005	5.0	250	0.8417	1.0641	0.8606	0.8417
2.2397	6.0	300	0.8779	0.8145	0.8996	0.8779
1.9058	7.0	350	0.9108	0.6322	0.9222	0.9108
1.7861	8.0	400	0.9404	0.5031	0.9477	0.9404
1.5948	9.0	450	0.9442	0.4105	0.9510	0.9442
1.5219	10.0	500	0.9546	0.3455	0.9613	0.9546
1.3415	11.0	550	0.9654	0.2813	0.9695	0.9654
1.2171	12.0	600	0.9629	0.2456	0.9671	0.9629
1.0959	13.0	650	0.975	0.1969	0.9771	0.975
1.0067	14.0	700	0.9792	0.1725	0.9809	0.9792
1.0844	15.0	750	0.9821	0.1507	0.9836	0.9821
0.931	16.0	800	0.9854	0.1307	0.9866	0.9854
0.8038	17.0	850	0.9858	0.1202	0.9870	0.9858
0.8623	18.0	900	0.9842	0.1083	0.9851	0.9842
0.7439	19.0	950	0.9879	0.0987	0.9891	0.9879
0.7537	20.0	1000	0.9892	0.0912	0.9903	0.9892
0.599	21.0	1050	0.9908	0.0788	0.9917	0.9908
0.6198	22.0	1100	0.9904	0.0711	0.9913	0.9904
0.5669	23.0	1150	0.9917	0.0663	0.9925	0.9917
0.5134	24.0	1200	0.9904	0.0630	0.9913	0.9904
0.5558	25.0	1250	0.99	0.0575	0.9909	0.99
0.5118	26.0	1300	0.9912	0.0589	0.9920	0.9912
0.5522	27.0	1350	0.9904	0.0517	0.9913	0.9904
0.4916	28.0	1400	0.9912	0.0487	0.9920	0.9912
0.3872	29.0	1450	0.9912	0.0440	0.9921	0.9912
0.4532	30.0	1500	0.9917	0.0464	0.9924	0.9917
0.4277	31.0	1550	0.9912	0.0408	0.9921	0.9912
0.4723	32.0	1600	0.9921	0.0378	0.9927	0.9921
0.3774	33.0	1650	0.9929	0.0351	0.9936	0.9929
0.3451	34.0	1700	0.9929	0.0368	0.9936	0.9929
0.3106	35.0	1750	0.9933	0.0349	0.9938	0.9933
0.2933	36.0	1800	0.9921	0.0364	0.9928	0.9921
0.2468	37.0	1850	0.9912	0.0369	0.9920	0.9912
0.461	38.0	1900	0.9921	0.0312	0.9928	0.9921
0.2706	39.0	1950	0.9933	0.0319	0.9939	0.9933
0.2784	40.0	2000	0.9925	0.0306	0.9932	0.9925
0.3167	41.0	2050	0.9929	0.0314	0.9936	0.9929
0.2242	42.0	2100	0.9929	0.0319	0.9936	0.9929
0.2439	43.0	2150	0.9929	0.0324	0.9937	0.9929
0.1995	44.0	2200	0.9938	0.0267	0.9943	0.9938
0.2178	45.0	2250	0.0273	0.9925	0.9932	0.9925
0.3018	46.0	2300	0.0281	0.9938	0.9943	0.9938
0.3096	47.0	2350	0.0285	0.9942	0.9948	0.9942
0.2636	48.0	2400	0.0261	0.9933	0.9941	0.9933
0.2441	49.0	2450	0.0233	0.9929	0.9937	0.9929
0.2102	50.0	2500	0.0255	0.9929	0.9935	0.9929
0.2302	51.0	2550	0.0268	0.9921	0.9928	0.9921
0.1548	52.0	2600	0.0251	0.9929	0.9935	0.9929
0.2293	53.0	2650	0.0264	0.9925	0.9932	0.9925
0.199	54.0	2700	0.0278	0.9942	0.9947	0.9942

Framework versions

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