You can deploy this model and start earning money today!

nik1509/
telugu_wav2vec_lr_ablation_5e-5

Name: telugu_wav2vec_lr_ablation_5e-5
Availability: InStock

transformers

safetensors

wav2vec2

automatic-speech-recognition

generated_from_trainer

nik1509/telugu_wav2vecmedium_basemodel

finetune

apache-2.0

endpoints_compatible

Views

No views yet

telugu_wav2vec_lr_ablation_5e-5

This model is a fine-tuned version of nik1509/telugu_wav2vecmedium_basemodel on the None dataset. It achieves the following results on the evaluation set:

Loss: 578.7206
Wer: 0.3931
Cer: 0.1677

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: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 15
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
342.1154	0.1736	200	451.8562	0.4218	0.1726
294.608	0.3472	400	457.1622	0.4158	0.1700
305.6544	0.5208	600	458.1221	0.4136	0.1701
330.206	0.6944	800	461.7636	0.4097	0.1691
320.7596	0.8681	1000	461.5178	0.4100	0.1691
331.7852	1.0417	1200	471.9964	0.4044	0.1675
303.8588	1.2153	1400	468.3314	0.4041	0.1682
246.6186	1.3889	1600	472.8593	0.4037	0.1676
308.5876	1.5625	1800	463.3277	0.4026	0.1683
286.4116	1.7361	2000	468.9177	0.4031	0.1678
318.4727	1.9097	2200	471.2498	0.4041	0.1679
257.4571	2.0833	2400	474.6279	0.4042	0.1677
319.8504	2.2569	2600	471.7175	0.4033	0.1675
271.5839	2.4306	2800	483.4916	0.3992	0.1667
300.9258	2.6042	3000	478.1400	0.3993	0.1671
330.3818	2.7778	3200	474.2046	0.3992	0.1677
257.3448	2.9514	3400	482.6506	0.3984	0.1672
252.6682	3.125	3600	481.7584	0.4007	0.1673
269.2817	3.2986	3800	487.3502	0.3987	0.1663
292.3175	3.4722	4000	488.7947	0.3989	0.1665
289.5761	3.6458	4200	480.8447	0.3989	0.1677
322.6325	3.8194	4400	481.7513	0.4015	0.1681
283.6295	3.9931	4600	479.9702	0.4051	0.1680
284.8582	4.1667	4800	483.8861	0.3981	0.1672
271.8823	4.3403	5000	484.5938	0.4012	0.1676
302.1243	4.5139	5200	493.2997	0.4006	0.1674
305.7237	4.6875	5400	492.8470	0.4022	0.1679
294.4779	4.8611	5600	500.1844	0.3969	0.1664
280.5959	5.0347	5800	494.2291	0.4013	0.1683
288.4263	5.2083	6000	506.2787	0.3975	0.1668
249.3606	5.3819	6200	507.8963	0.3949	0.1667
278.6182	5.5556	6400	499.4547	0.3977	0.1675
267.5538	5.7292	6600	515.1586	0.3945	0.1662
297.6201	5.9028	6800	505.4439	0.3978	0.1671
241.6456	6.0764	7000	518.3139	0.3941	0.1660
301.2793	6.25	7200	504.7360	0.3964	0.1669
280.7293	6.4236	7400	507.5990	0.3997	0.1682
259.4739	6.5972	7600	512.6235	0.3954	0.1667
335.192	6.7708	7800	510.1955	0.3953	0.1667
267.8126	6.9444	8000	514.3646	0.3943	0.1665
289.4583	7.1181	8200	525.1478	0.3942	0.1661
289.1285	7.2917	8400	525.4915	0.3923	0.1666
224.3789	7.4653	8600	518.1742	0.3973	0.1677
260.6031	7.6389	8800	517.0095	0.3962	0.1669
290.3712	7.8125	9000	521.8460	0.3985	0.1681
231.2147	7.9861	9200	527.4080	0.3950	0.1660
261.6565	8.1597	9400	525.9984	0.3990	0.1673
226.5641	8.3333	9600	522.2488	0.3991	0.1683
290.2888	8.5069	9800	538.3866	0.3925	0.1665
247.3912	8.6806	10000	535.1982	0.3904	0.1671
265.4921	8.8542	10200	532.2893	0.3949	0.1672
259.7409	9.0278	10400	525.2161	0.3962	0.1675
220.7795	9.2014	10600	543.8533	0.3921	0.1666
277.737	9.375	10800	548.3875	0.3922	0.1670
227.5874	9.5486	11000	536.6715	0.3936	0.1670
286.3873	9.7222	11200	539.8420	0.3955	0.1676
257.8596	9.8958	11400	546.0359	0.3937	0.1663
243.5228	10.0694	11600	556.4694	0.3916	0.1666
246.7866	10.2431	11800	542.3545	0.3937	0.1680
266.3281	10.4167	12000	548.3701	0.3933	0.1672
249.7809	10.5903	12200	534.3046	0.3942	0.1679
271.9256	10.7639	12400	543.2160	0.3928	0.1669
226.6287	10.9375	12600	542.5263	0.3956	0.1676
229.5597	11.1111	12800	551.5705	0.3951	0.1676
250.9209	11.2847	13000	553.2479	0.3942	0.1671
230.6768	11.4583	13200	547.4739	0.3972	0.1683
225.6195	11.6319	13400	547.4948	0.3941	0.1681
268.5525	11.8056	13600	558.0397	0.3960	0.1678
233.5317	11.9792	13800	552.9329	0.3929	0.1679
228.7924	12.1528	14000	558.8180	0.3943	0.1677
238.6032	12.3264	14200	567.7181	0.3952	0.1676
236.1686	12.5	14400	564.1912	0.3941	0.1679
230.2658	12.6736	14600	564.6769	0.3964	0.1677
222.8885	12.8472	14800	571.9883	0.3928	0.1673
214.2647	13.0208	15000	571.7230	0.3960	0.1678
253.5084	13.1944	15200	577.7383	0.3957	0.1683
252.1588	13.3681	15400	570.7164	0.3936	0.1687
211.601	13.5417	15600	569.6709	0.3951	0.1682
223.5046	13.7153	15800	565.7266	0.3925	0.1682
222.2304	13.8889	16000	566.3947	0.3935	0.1686
228.0174	14.0625	16200	581.0183	0.3889	0.1676
205.2938	14.2361	16400	581.5366	0.3936	0.1677
231.1113	14.4097	16600	577.0634	0.3906	0.1677
242.2721	14.5833	16800	568.3370	0.3934	0.1685
241.0949	14.7569	17000	577.6273	0.3915	0.1678
229.9881	14.9306	17200	578.7206	0.3931	0.1677

Framework versions

Transformers 4.52.4
Pytorch 2.8.0.dev20250609+cu118
Datasets 3.6.0
Tokenizers 0.21.1

nik1509/telugu_wav2vec_lr_ablation_5e-5

telugu_wav2vec_lr_ablation_5e-5

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

nik1509/
telugu_wav2vec_lr_ablation_5e-5