You can deploy this model and start earning money today!

rohitp1/
libri-smallw2v2-no-copy-kl-alpha-0.75-T-1-take-3

Name: libri-smallw2v2-no-copy-kl-alpha-0.75-T-1-take-3
Availability: InStock

transformers

pytorch

wav2vec2

automatic-speech-recognition

generated_from_trainer

endpoints_compatible

Views

No views yet

libri-smallw2v2-no-copy-kl-alpha-0.75-T-1-take-3

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

Loss: 233.4488
Wer: 0.3266

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: 32
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 64
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.2
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
344.6087	1.12	400	259.4728	0.4553
331.2433	2.24	800	249.7617	0.4505
324.318	3.36	1200	243.3380	0.4476
318.8272	4.48	1600	245.5005	0.4397
314.193	5.6	2000	245.8428	0.4389
310.3951	6.72	2400	249.0932	0.4358
307.6068	7.84	2800	248.7932	0.4386
308.3191	8.96	3200	242.1969	0.4309
302.0043	10.08	3600	251.7498	0.4271
296.6045	11.2	4000	246.3982	0.4272
295.0746	12.32	4400	249.5031	0.4201
294.2732	13.45	4800	247.5096	0.4167
290.1665	14.57	5200	247.6531	0.4227
293.2169	15.69	5600	246.6296	0.4150
287.6487	16.81	6000	244.2763	0.4132
287.8109	17.93	6400	243.7672	0.4116
282.7126	19.05	6800	241.8889	0.4073
280.5111	20.17	7200	251.7473	0.4015
276.0679	21.29	7600	242.1010	0.3990
275.7184	22.41	8000	244.3330	0.3966
273.0371	23.53	8400	240.5063	0.3908
268.2875	24.65	8800	241.2827	0.3916
262.7938	25.77	9200	236.9669	0.3870
262.0252	26.89	9600	238.0447	0.3837
256.3527	28.01	10000	231.9627	0.3777
253.4488	29.13	10400	241.3886	0.3778
251.0047	30.25	10800	239.3254	0.3716
247.1357	31.37	11200	234.5317	0.3743
245.7466	32.49	11600	237.2660	0.3732
240.7763	33.61	12000	234.2133	0.3719
240.4568	34.73	12400	233.6447	0.3652
236.5008	35.85	12800	231.5317	0.3634
234.3844	36.97	13200	235.9599	0.3667
234.8658	38.1	13600	235.8033	0.3629
229.0455	39.22	14000	235.2613	0.3597
228.1712	40.34	14400	237.7952	0.3559
225.4442	41.46	14800	237.6732	0.3553
223.492	42.58	15200	228.6896	0.3549
222.2095	43.7	15600	233.7846	0.3528
221.3752	44.82	16000	235.6401	0.3503
220.0048	45.94	16400	236.2913	0.3486
214.734	47.06	16800	233.7592	0.3452
213.6554	48.18	17200	233.3319	0.3468
212.7388	49.3	17600	232.7798	0.3447
210.9421	50.42	18000	239.8152	0.3483
211.6293	51.54	18400	235.0050	0.3450
209.8978	52.66	18800	235.3156	0.3453
207.996	53.78	19200	233.3227	0.3429
206.4369	54.9	19600	231.2948	0.3395
202.3726	56.02	20000	226.5554	0.3387
201.5557	57.14	20400	236.6525	0.3413
203.2557	58.26	20800	231.1979	0.3387
200.6613	59.38	21200	232.5989	0.3361
199.9518	60.5	21600	233.6743	0.3368
198.7427	61.62	22000	236.3777	0.3365
195.5101	62.75	22400	230.0184	0.3354
195.4992	63.87	22800	229.9954	0.3332
193.023	64.99	23200	233.4538	0.3343
195.5294	66.11	23600	235.9190	0.3328
193.4176	67.23	24000	235.0465	0.3316
191.4303	68.35	24400	234.2414	0.3331
190.1889	69.47	24800	234.3200	0.3306
188.0727	70.59	25200	231.8731	0.3288
189.5906	71.71	25600	234.2662	0.3297
187.4333	72.83	26000	234.4295	0.3298
188.8704	73.95	26400	234.3622	0.3272
186.7061	75.07	26800	233.1743	0.3257
185.7288	76.19	27200	233.8410	0.3255
184.4937	77.31	27600	230.8933	0.3249
183.4367	78.43	28000	233.5291	0.3274
184.048	79.55	28400	233.0128	0.3274
184.0794	80.67	28800	232.1064	0.3259
181.9872	81.79	29200	234.1832	0.3259
181.738	82.91	29600	233.4488	0.3266

Framework versions

Transformers 4.25.1
Pytorch 1.12.1
Datasets 2.7.1
Tokenizers 0.11.0

rohitp1/libri-smallw2v2-no-copy-kl-alpha-0.75-T-1-take-3

libri-smallw2v2-no-copy-kl-alpha-0.75-T-1-take-3

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

rohitp1/
libri-smallw2v2-no-copy-kl-alpha-0.75-T-1-take-3