indicwav2vec_outputs

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: nan
• Cer: 1.0
• Wer: 1.0

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 1011
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• num_epochs: 35.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
3.8795	0.3028	500	3.7869	0.9860	1.0
1.8805	0.6057	1000	2.0423	0.4124	0.6416
1.5823	0.9085	1500	1.7622	0.3701	0.5792
2.2702	1.2114	2000	2.0595	0.5233	0.8442
2.7429	1.5142	2500	2.9181	0.8706	0.9792
3.1077	1.8171	3000	3.0393	0.9061	0.9898
2.9896	2.1199	3500	2.8581	0.8528	0.9778
3.2643	2.4228	4000	3.0456	0.8025	0.9649
3.6542	2.7256	4500	3.4606	0.8008	0.9658
3.7622	3.0285	5000	3.6476	0.8315	0.9835
3.8614	3.3313	5500	3.8326	0.8628	0.9924
3.9769	3.6342	6000	4.0055	0.8808	0.9953
4.1241	3.9370	6500	4.1374	0.8920	0.9965
4.1261	4.2399	7000	4.1374	0.8920	0.9965
4.1009	4.5427	7500	4.1374	0.8920	0.9965
4.1698	4.8455	8000	4.1374	0.8920	0.9965
4.129	5.1484	8500	4.1374	0.8920	0.9965
4.1413	5.4512	9000	4.1374	0.8920	0.9965
4.122	5.7541	9500	4.1374	0.8920	0.9965
4.1652	6.0569	10000	4.1374	0.8920	0.9965
4.1801	6.3598	10500	4.1374	0.8920	0.9965
4.092	6.6626	11000	4.1374	0.8920	0.9965
4.0204	6.9655	11500	4.1374	0.8920	0.9965
4.1036	7.2683	12000	4.1374	0.8920	0.9965
4.1918	7.5712	12500	4.1374	0.8920	0.9965
4.1059	7.8740	13000	4.1374	0.8920	0.9965
4.0833	8.1769	13500	4.1374	0.8920	0.9965
4.1278	8.4797	14000	4.1374	0.8920	0.9965
4.1365	8.7826	14500	4.1374	0.8920	0.9965
4.1201	9.0854	15000	4.1374	0.8920	0.9965
4.1476	9.3882	15500	4.1374	0.8920	0.9965
4.0935	9.6911	16000	4.1374	0.8920	0.9965
4.1109	9.9939	16500	4.1374	0.8920	0.9965
4.1389	10.2968	17000	4.1374	0.8920	0.9965
4.0907	10.5996	17500	4.1374	0.8920	0.9965
4.0825	10.9025	18000	4.1374	0.8920	0.9965
4.1094	11.2053	18500	4.1374	0.8920	0.9965
4.0689	11.5082	19000	4.1374	0.8920	0.9965
4.0984	11.8110	19500	4.1374	0.8920	0.9965
4.0569	12.1139	20000	4.1374	0.8920	0.9965
4.1462	12.4167	20500	4.1374	0.8920	0.9965
4.1554	12.7196	21000	4.1374	0.8920	0.9965
4.2207	13.0224	21500	4.1374	0.8920	0.9965
4.1518	13.3253	22000	4.1374	0.8920	0.9965
4.1521	13.6281	22500	4.1374	0.8920	0.9965
4.1367	13.9310	23000	4.1374	0.8920	0.9965
4.0904	14.2338	23500	4.1374	0.8920	0.9965
4.0813	14.5366	24000	4.1374	0.8920	0.9965
4.1001	14.8395	24500	4.1374	0.8920	0.9965
4.1333	15.1423	25000	4.1374	0.8920	0.9965
4.0785	15.4452	25500	4.1374	0.8920	0.9965
4.1651	15.7480	26000	4.1374	0.8920	0.9965
4.0987	16.0509	26500	4.1374	0.8920	0.9965
4.1327	16.3537	27000	4.1374	0.8920	0.9965
4.1128	16.6566	27500	4.1374	0.8920	0.9965
4.0694	16.9594	28000	4.1374	0.8920	0.9965
5.946	17.2623	28500	nan	1.0	1.0
0.0	17.5651	29000	nan	1.0	1.0
0.0	17.8680	29500	nan	1.0	1.0
0.0	18.1708	30000	nan	1.0	1.0
0.0	18.4737	30500	nan	1.0	1.0
0.0	18.7765	31000	nan	1.0	1.0
0.0	19.0793	31500	nan	1.0	1.0
0.0	19.3822	32000	nan	1.0	1.0
0.0	19.6850	32500	nan	1.0	1.0
0.0	19.9879	33000	nan	1.0	1.0
0.0	20.2907	33500	nan	1.0	1.0
0.0	20.5936	34000	nan	1.0	1.0
0.0	20.8964	34500	nan	1.0	1.0
0.0	21.1993	35000	nan	1.0	1.0
0.0	21.5021	35500	nan	1.0	1.0
0.0	21.8050	36000	nan	1.0	1.0
0.0	22.1078	36500	nan	1.0	1.0
0.0	22.4107	37000	nan	1.0	1.0
0.0	22.7135	37500	nan	1.0	1.0
0.0	23.0164	38000	nan	1.0	1.0
0.0	23.3192	38500	nan	1.0	1.0
0.0	23.6220	39000	nan	1.0	1.0
0.0	23.9249	39500	nan	1.0	1.0
0.0	24.2277	40000	nan	1.0	1.0
0.0	24.5306	40500	nan	1.0	1.0
0.0	24.8334	41000	nan	1.0	1.0
0.0	25.1363	41500	nan	1.0	1.0
0.0	25.4391	42000	nan	1.0	1.0
0.0	25.7420	42500	nan	1.0	1.0
0.0	26.0448	43000	nan	1.0	1.0
0.0	26.3477	43500	nan	1.0	1.0
0.0	26.6505	44000	nan	1.0	1.0
0.0	26.9534	44500	nan	1.0	1.0
0.0	27.2562	45000	nan	1.0	1.0
0.0	27.5591	45500	nan	1.0	1.0
0.0	27.8619	46000	nan	1.0	1.0
0.0	28.1647	46500	nan	1.0	1.0
0.0	28.4676	47000	nan	1.0	1.0
0.0	28.7704	47500	nan	1.0	1.0
0.0	29.0733	48000	nan	1.0	1.0
0.0	29.3761	48500	nan	1.0	1.0
0.0	29.6790	49000	nan	1.0	1.0
0.0	29.9818	49500	nan	1.0	1.0
0.0	30.2847	50000	nan	1.0	1.0
0.0	30.5875	50500	nan	1.0	1.0
0.0	30.8904	51000	nan	1.0	1.0
0.0	31.1932	51500	nan	1.0	1.0
0.0	31.4961	52000	nan	1.0	1.0
0.0	31.7989	52500	nan	1.0	1.0
0.0	32.1018	53000	nan	1.0	1.0
0.0	32.4046	53500	nan	1.0	1.0
0.0	32.7075	54000	nan	1.0	1.0
0.0	33.0103	54500	nan	1.0	1.0
0.0	33.3131	55000	nan	1.0	1.0
0.0	33.6160	55500	nan	1.0	1.0
0.0	33.9188	56000	nan	1.0	1.0
0.0	34.2217	56500	nan	1.0	1.0
0.0	34.5245	57000	nan	1.0	1.0
0.0	34.8274	57500	nan	1.0	1.0

Framework versions

• Transformers 4.43.1
• Pytorch 2.4.0
• Datasets 2.20.0
• Tokenizers 0.19.1