assis

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8484
• Wer: 1

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 3000
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
22.3611	0.45	100	25.5854	1
20.8274	0.9	200	20.4977	1
12.1089	1.35	300	11.0220	1
5.043	1.81	400	4.3838	1
3.788	2.26	500	3.5831	1
3.445	2.71	600	3.4112	1
3.3042	3.16	700	3.3104	1
3.221	3.61	800	3.2255	1
3.1628	4.06	900	3.1618	1
3.0645	4.51	1000	3.1010	1
3.0913	4.97	1100	3.0624	1
3.0819	5.42	1200	3.0136	1
2.9502	5.87	1300	2.9883	1
2.9611	6.32	1400	2.9651	1
2.9287	6.77	1500	2.9474	1
2.9461	7.22	1600	2.9280	1
2.9176	7.67	1700	2.9148	1
2.8986	8.13	1800	2.9138	1
2.8896	8.58	1900	2.9050	1
2.8879	9.03	2000	2.9093	1
2.9085	9.48	2100	2.8998	1
2.876	9.93	2200	2.8807	1
2.8649	10.38	2300	2.8734	1
2.8653	10.84	2400	2.8681	1
2.8683	11.29	2500	2.8596	1
2.8452	11.74	2600	2.8667	1
2.8468	12.19	2700	2.8514	1
2.846	12.64	2800	2.8541	1
2.8415	13.09	2900	2.8493	1
2.8195	13.54	3000	2.8472	1
2.8103	14.0	3100	2.8244	1
2.6495	14.45	3200	2.5809	1
2.3126	14.9	3300	2.1612	1
1.92	15.35	3400	1.7312	1
1.5734	15.8	3500	1.4245	1
1.4081	16.25	3600	1.2659	1
1.2573	16.7	3700	1.1694	1
1.194	17.16	3800	1.0930	1
1.1053	17.61	3900	1.0393	1
1.072	18.06	4000	0.9792	1
1.0148	18.51	4100	0.9468	1
0.9995	18.96	4200	0.9228	1
0.9688	19.41	4300	0.9071	1
0.956	19.86	4400	0.8950	1
0.9565	20.32	4500	0.8632	1
0.9215	20.77	4600	0.8673	1
0.9006	21.22	4700	0.8647	1
0.8645	21.67	4800	0.8566	1
0.8768	22.12	4900	0.8527	1
0.8809	22.57	5000	0.8484	1

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3