You can deploy this model and start earning money today!

Samuael/
asr-amharic-phoneme-based-37-4

Name: asr-amharic-phoneme-based-37-4
Availability: InStock

automatic-speech-recognition

generated_from_trainer

Samuael/asr-amharic-phoneme-based-37-4

finetune

apache-2.0

endpoints_compatible

Views

No views yet

asr-amharic-phoneme-based-37-4

This model is a fine-tuned version of Samuael/asr-amharic-phoneme-based-37-4 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.2375
Wer: 0.3633
Phoneme Cer: 0.0675
Cer: 0.1103

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: 5e-05
train_batch_size: 32
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 100
num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Phoneme Cer	Cer
3.1066	0.59	200	3.2635	1.0	1.0	1.0
3.0018	1.18	400	3.2424	1.0	1.0	1.0
2.9877	1.76	600	3.0748	1.0	1.0	1.0
2.9609	2.35	800	3.0010	1.0	1.0	1.0
2.4744	2.94	1000	2.3659	1.0	0.9150	1.0000
1.221	3.53	1200	0.8924	0.8893	0.2290	0.3507
0.7713	4.12	1400	0.5988	0.6852	0.1544	0.2408
0.7683	4.71	1600	0.4776	0.6150	0.1299	0.2020
0.7059	5.29	1800	0.4349	0.5580	0.1164	0.1819
0.7168	5.88	2000	0.3947	0.5505	0.1106	0.1738
0.5871	6.47	2200	0.3703	0.5510	0.1078	0.1696
0.5302	7.06	2400	0.3443	0.5227	0.1015	0.1609
0.5622	7.65	2600	0.3325	0.4937	0.0967	0.1535
0.4664	8.24	2800	0.3216	0.4967	0.0961	0.1518
0.4083	8.82	3000	0.3119	0.4806	0.0926	0.1475
0.4234	9.41	3200	0.3117	0.4678	0.0896	0.1423
0.4318	10.0	3400	0.2938	0.4637	0.0881	0.1398
0.6403	10.59	3600	0.2960	0.4402	0.0855	0.1363
0.4015	11.18	3800	0.2955	0.4605	0.0868	0.1381
0.3722	11.76	4000	0.2806	0.4390	0.0832	0.1329
0.3153	12.35	4200	0.2801	0.4201	0.0817	0.1302
0.3262	12.94	4400	0.2681	0.4316	0.0816	0.1311
0.3238	13.53	4600	0.2685	0.4136	0.0786	0.1262
0.311	14.12	4800	0.2671	0.4018	0.0769	0.1231
0.3581	14.71	5000	0.2651	0.3968	0.0765	0.1228
0.3355	15.29	5200	0.2634	0.3970	0.0762	0.1225
0.5307	15.88	5400	0.2632	0.4007	0.0760	0.1226
0.3148	16.47	5600	0.2579	0.3964	0.0745	0.1202
0.3247	17.06	5800	0.2562	0.3825	0.0726	0.1176
0.3346	17.65	6000	0.2498	0.3895	0.0733	0.1183
0.2601	18.24	6200	0.2512	0.3924	0.0725	0.1176
0.2705	18.82	6400	0.2459	0.4010	0.0737	0.1198
0.2295	19.41	6600	0.2492	0.3826	0.0716	0.1159
0.3824	20.0	6800	0.2532	0.3820	0.0720	0.1163
0.3082	20.59	7000	0.2485	0.3796	0.0714	0.1157
0.2927	21.18	7200	0.2477	0.3807	0.0712	0.1155
0.3819	21.76	7400	0.2446	0.3729	0.0700	0.1139
0.2941	22.35	7600	0.2452	0.3740	0.0699	0.1138
0.3104	22.94	7800	0.2406	0.3699	0.0688	0.1126
0.2359	23.53	8000	0.2389	0.3731	0.0692	0.1129
0.2895	24.12	8200	0.2405	0.3613	0.0682	0.1114
0.2144	24.71	8400	0.2404	0.3686	0.0682	0.1115
0.3572	25.29	8600	0.2376	0.3769	0.0693	0.1133
0.3137	25.88	8800	0.2383	0.3660	0.0685	0.1115
0.2457	26.47	9000	0.2379	0.3688	0.0684	0.1115
0.2577	27.06	9200	0.2369	0.3707	0.0682	0.1119
0.3276	27.65	9400	0.2379	0.3621	0.0676	0.1105
0.2566	28.24	9600	0.2376	0.3637	0.0672	0.1104
0.2814	28.82	9800	0.2369	0.3652	0.0676	0.1107
0.2574	29.41	10000	0.2372	0.3614	0.0673	0.1102
0.5213	30.0	10200	0.2375	0.3633	0.0675	0.1103

Framework versions

Transformers 4.35.2
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.0