ssc-el-CY-model

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.7970
• Cer: 0.9445
• Wer: 0.9999

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
6.6563	0.1883	100	3.5095	0.9958	1.0
3.5193	0.3766	200	3.5593	0.9958	1.0
3.9095	0.5650	300	3.3924	0.9958	1.0
3.6929	0.7533	400	3.3523	0.9958	1.0
3.5347	0.9416	500	3.4397	0.9958	1.0
3.48	1.1299	600	3.3551	0.9958	1.0
3.5095	1.3183	700	3.3840	0.9958	1.0
3.4933	1.5066	800	3.3772	0.9958	1.0
3.4675	1.6949	900	3.4572	0.9958	1.0
3.4735	1.8832	1000	3.4976	0.9843	0.9995
3.4659	2.0716	1100	3.2917	0.9922	1.0
3.4585	2.2599	1200	3.2933	0.9915	0.9829
3.4264	2.4482	1300	3.3039	0.9915	0.9830
3.4353	2.6365	1400	3.2864	0.9914	0.9830
3.4617	2.8249	1500	3.2812	0.9916	0.9834
3.4312	3.0132	1600	3.3325	0.9915	0.9830
3.3992	3.2015	1700	3.2775	0.9915	0.9833
3.3732	3.3898	1800	3.4619	0.9857	0.9998
3.3619	3.5782	1900	3.4054	0.9916	0.9900
3.3602	3.7665	2000	3.3870	0.9911	0.9870
3.3476	3.9548	2100	3.3257	0.9914	0.9838
3.3112	4.1431	2200	3.4524	0.9901	0.9891
3.3166	4.3315	2300	3.3271	0.9913	0.9859
3.3059	4.5198	2400	3.4313	0.9914	0.9835
3.2864	4.7081	2500	3.2850	0.9852	0.9944
3.2992	4.8964	2600	3.3862	0.9895	0.9885
3.2491	5.0847	2700	3.3486	0.9914	0.9831
3.2507	5.2731	2800	3.2751	0.9911	0.9841
3.1689	5.4614	2900	3.3519	0.9852	0.9994
3.1484	5.6497	3000	3.3401	0.9841	0.9944
3.0872	5.8380	3100	3.1924	0.9804	0.9976
2.9885	6.0264	3200	3.0345	0.9180	1.0
2.9312	6.2147	3300	2.9692	0.8770	1.0
2.8406	6.4030	3400	2.9859	0.8145	1.0
2.7876	6.5913	3500	2.8495	0.8633	0.9999
2.7776	6.7797	3600	2.8264	0.8823	1.0
2.6837	6.9680	3700	2.8850	0.7760	0.9999
2.6025	7.1563	3800	2.7891	0.7636	0.9982
2.5765	7.3446	3900	2.7292	0.8369	1.0
2.5398	7.5330	4000	2.8252	0.7108	0.9896
2.5163	7.7213	4100	2.7143	0.7544	0.9988
2.4676	7.9096	4200	2.6224	0.7611	0.9976
2.4589	8.0979	4300	2.7647	0.7194	0.9957
2.4198	8.2863	4400	2.6359	0.7458	0.9981
2.3952	8.4746	4500	2.6756	0.7132	0.9939
2.3696	8.6629	4600	2.6740	0.7052	0.9911
2.3592	8.8512	4700	2.6290	0.7753	1.0
2.2945	9.0395	4800	2.5619	0.7157	0.9953
2.3084	9.2279	4900	2.6623	0.7370	0.9991
2.3028	9.4162	5000	2.6039	0.6887	0.9909
2.2883	9.6045	5100	2.6003	0.7232	0.9965
2.3452	9.7928	5200	2.5988	0.7459	0.9964
2.305	9.9812	5300	2.6939	0.7314	0.9973
2.3447	10.1695	5400	2.6968	0.7463	0.9994
2.3686	10.3578	5500	2.6992	0.7833	0.9993
2.4429	10.5461	5600	2.6674	0.7284	0.9991
2.4139	10.7345	5700	2.6574	0.7682	0.9988
2.405	10.9228	5800	2.6806	0.7395	0.9999
2.4375	11.1111	5900	2.6505	0.7145	0.9993
2.4065	11.2994	6000	2.8557	0.7743	0.9999
2.5238	11.4878	6100	2.8419	0.7982	0.9999
2.6059	11.6761	6200	2.7876	0.8675	1.0
2.5852	11.8644	6300	2.8252	0.8875	1.0
2.5776	12.0527	6400	2.7428	0.9291	1.0
2.5932	12.2411	6500	2.7696	0.8976	0.9999
2.6125	12.4294	6600	2.7940	0.9234	1.0
2.6714	12.6177	6700	2.7963	0.9418	1.0
2.6844	12.8060	6800	2.9210	0.9862	0.9995
2.7054	12.9944	6900	2.8712	0.9838	0.9990
2.6509	13.1827	7000	2.9206	0.9718	0.9998
2.7351	13.3710	7100	2.9365	0.9731	0.9995
2.7265	13.5593	7200	2.9177	0.9943	0.9984
2.817	13.7476	7300	2.9292	0.9881	0.9972
2.7506	13.9360	7400	2.9558	0.9909	0.9979
2.799	14.1243	7500	2.9085	0.9486	0.9998
2.7322	14.3126	7600	2.8605	0.9283	1.0
2.6716	14.5009	7700	2.8886	0.8909	1.0
2.6592	14.6893	7800	2.8992	0.8688	1.0
2.6646	14.8776	7900	2.9344	0.8686	1.0
2.724	15.0659	8000	2.9167	0.9103	1.0
2.8567	15.2542	8100	2.9886	0.9291	1.0
3.3181	15.4426	8200	3.2414	0.9603	0.9999
3.6228	15.6309	8300	4.0603	0.9935	0.9999
3.2903	15.8192	8400	3.1761	0.9610	0.9999
3.3319	16.0075	8500	3.1989	0.9701	0.9996
3.1339	16.1959	8600	3.4163	0.9837	0.9997
3.1833	16.3842	8700	3.2594	0.9772	0.9996
3.1457	16.5725	8800	3.2544	0.9769	0.9995
3.1402	16.7608	8900	3.2880	0.9776	0.9993
3.1477	16.9492	9000	3.2459	0.9755	0.9994
3.0863	17.1375	9100	3.2432	0.9612	0.9997
3.1396	17.3258	9200	3.2105	0.9637	0.9997
3.0992	17.5141	9300	3.2458	0.9566	0.9994
3.1286	17.7024	9400	3.2120	0.9605	0.9995
3.1312	17.8908	9500	3.2466	0.9570	0.9994
3.1014	18.0791	9600	3.2039	0.9668	0.9995
3.1079	18.2674	9700	3.2022	0.9632	0.9997
3.1021	18.4557	9800	3.2071	0.9608	0.9996
3.0741	18.6441	9900	3.2062	0.9593	0.9993
3.0982	18.8324	10000	3.2004	0.9591	0.9995
3.1016	19.0207	10100	3.2120	0.9552	0.9995
3.0939	19.2090	10200	3.2339	0.9659	0.9997
3.0702	19.3974	10300	3.2106	0.9597	0.9998
3.0876	19.5857	10400	3.1958	0.9546	0.9996
3.091	19.7740	10500	3.2048	0.9605	0.9996
3.0678	19.9623	10600	3.2073	0.9614	0.9997
3.0627	20.1507	10700	3.2349	0.9421	0.9988
3.115	20.3390	10800	3.2771	0.9384	0.9984
3.0859	20.5273	10900	3.2839	0.9350	0.9979
3.0783	20.7156	11000	3.2618	0.9380	0.9984
3.063	20.9040	11100	3.2845	0.9318	0.9968
3.1253	21.0923	11200	3.2100	0.9575	0.9996
3.1022	21.2806	11300	3.2077	0.9570	0.9994
3.0577	21.4689	11400	3.2107	0.9566	0.9991
3.1657	21.6573	11500	3.2189	0.9536	0.9994
3.104	21.8456	11600	3.2216	0.9526	0.9995
3.1261	22.0339	11700	3.2874	0.9331	0.9979
3.1522	22.2222	11800	3.2801	0.9351	0.9982
3.1181	22.4105	11900	3.2882	0.9340	0.9980
3.1137	22.5989	12000	3.3047	0.9278	0.9970
3.1867	22.7872	12100	3.3426	0.9236	0.9974
3.2373	22.9755	12200	3.3194	0.9272	0.9976
3.1979	23.1638	12300	3.3497	0.9263	0.9977
3.2752	23.3522	12400	3.3670	0.9256	0.9978
3.2831	23.5405	12500	3.3733	0.9209	0.9973
3.1949	23.7288	12600	3.4054	0.9129	0.9963
3.2813	23.9171	12700	3.4314	0.9127	0.9968
3.2636	24.1055	12800	3.3522	0.9395	0.9992
3.3405	24.2938	12900	3.3866	0.9271	0.9981
3.3643	24.4821	13000	3.3755	0.9349	0.9988
3.3405	24.6704	13100	3.4095	0.9257	0.9982
3.3849	24.8588	13200	3.4442	0.9180	0.9976
3.4133	25.0471	13300	3.5845	0.8960	0.9952
3.4747	25.2354	13400	3.5972	0.8955	0.9950
3.3754	25.4237	13500	3.5783	0.9000	0.9954
3.6186	25.6121	13600	3.6207	0.8970	0.9952
3.5257	25.8004	13700	3.5450	0.9204	0.9982
3.5552	25.9887	13800	3.5970	0.9126	0.9978
3.5488	26.1770	13900	3.5955	0.9252	0.9987
3.7153	26.3653	14000	3.6388	0.9249	0.9990
3.5743	26.5537	14100	3.6605	0.9255	0.9993
3.5399	26.7420	14200	3.6604	0.9284	0.9993
3.5299	26.9303	14300	3.6808	0.9279	0.9994
3.5682	27.1186	14400	3.6842	0.9275	0.9993
3.6293	27.3070	14500	3.6912	0.9284	0.9994
3.7195	27.4953	14600	3.6877	0.9300	0.9993
3.7357	27.6836	14700	3.6994	0.9340	0.9994
3.6613	27.8719	14800	3.6993	0.9319	0.9995
3.6398	28.0603	14900	3.7152	0.9317	0.9994
3.7112	28.2486	15000	3.7308	0.9372	0.9999
3.6608	28.4369	15100	3.7434	0.9404	0.9999
3.7395	28.6252	15200	3.7598	0.9421	0.9999
3.7235	28.8136	15300	3.7651	0.9412	0.9999
3.6748	29.0019	15400	3.7793	0.9431	0.9999
3.7072	29.1902	15500	3.7839	0.9440	0.9999
3.7432	29.3785	15600	3.7883	0.9437	0.9999
3.692	29.5669	15700	3.7917	0.9438	0.9999
3.8549	29.7552	15800	3.7958	0.9445	1.0
3.6699	29.9435	15900	3.7970	0.9445	0.9999

Framework versions

• Transformers 4.57.2
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.22.0