wav2vec-large-en

This model is a fine-tuned version of facebook/wav2vec2-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.8842
• 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: 5e-06
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• 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
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.3521	50	6.2209	1.0
7.7114	0.7042	100	5.8125	1.0
7.7114	1.0563	150	5.5788	1.0
6.2512	1.4085	200	5.5017	1.0
6.2512	1.7606	250	5.0778	1.0
5.152	2.1127	300	3.0842	1.0
5.152	2.4648	350	2.9082	1.0
2.9862	2.8169	400	2.9040	1.0
2.9862	3.1690	450	2.9172	1.0
2.9346	3.5211	500	3.0595	1.0
2.9346	3.8732	550	2.8957	1.0
2.9037	4.2254	600	2.9380	1.0
2.9037	4.5775	650	2.8769	1.0
2.8894	4.9296	700	2.9011	1.0
2.8894	5.2817	750	2.8993	1.0
2.8886	5.6338	800	2.8978	1.0
2.8886	5.9859	850	2.8762	1.0
2.8831	6.3380	900	2.8726	1.0
2.8831	6.6901	950	2.8747	1.0
2.8772	7.0423	1000	3.0238	1.0
2.8772	7.3944	1050	2.8764	1.0
2.8913	7.7465	1100	2.8801	1.0
2.8913	8.0986	1150	2.8926	1.0
2.8677	8.4507	1200	2.9241	1.0
2.8677	8.8028	1250	2.9400	1.0
2.8713	9.1549	1300	2.9158	1.0
2.8713	9.5070	1350	2.8834	1.0
2.8702	9.8592	1400	2.8683	1.0
2.8702	10.2113	1450	2.8976	1.0
2.8817	10.5634	1500	2.9263	1.0
2.8817	10.9155	1550	2.8732	1.0
2.8657	11.2676	1600	2.9270	1.0
2.8657	11.6197	1650	2.8860	1.0
2.8618	11.9718	1700	2.8889	1.0
2.8618	12.3239	1750	2.8942	1.0
2.8846	12.6761	1800	2.8856	1.0
2.8846	13.0282	1850	2.9049	1.0
2.8635	13.3803	1900	2.8727	1.0
2.8635	13.7324	1950	2.8900	1.0
2.8634	14.0845	2000	2.9005	1.0
2.8634	14.4366	2050	2.9035	1.0
2.8624	14.7887	2100	2.9030	1.0
2.8624	15.1408	2150	2.8754	1.0
2.8627	15.4930	2200	2.8944	1.0
2.8627	15.8451	2250	2.8651	1.0
2.8595	16.1972	2300	2.9012	1.0
2.8595	16.5493	2350	2.9074	1.0
2.8591	16.9014	2400	2.8764	1.0
2.8591	17.2535	2450	2.8917	1.0
2.8615	17.6056	2500	2.8757	1.0
2.8615	17.9577	2550	2.8945	1.0
2.861	18.3099	2600	2.9263	1.0
2.861	18.6620	2650	2.8842	1.0

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Tokenizers 0.20.3