wav2vec2-ft-lre5-adm-ga2b16-st30k-v2

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

• Loss: 3.2472
• Wer: 1.0000
• Cer: 1.0000

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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_steps: 500
• training_steps: 30000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.4116	0.4165	1000	3.3778	1.0000	1.0000
3.2898	0.8330	2000	3.3594	1.0000	1.0000
3.2155	1.2495	3000	3.2941	1.0000	1.0000
3.2842	1.6660	4000	3.3287	1.0000	1.0000
3.2136	2.0825	5000	3.2953	1.0000	1.0000
3.2529	2.4990	6000	3.2965	1.0000	1.0000
3.2371	2.9155	7000	3.2881	1.0000	1.0000
3.2658	3.3319	8000	3.2694	1.0000	1.0000
3.1706	3.7484	9000	3.2601	1.0000	1.0000
3.1812	4.1649	10000	3.2748	1.0000	1.0000
3.1664	4.5814	11000	3.2633	1.0000	1.0000
3.1596	4.9979	12000	3.2687	1.0000	1.0000
3.1412	5.4144	13000	3.2663	1.0000	1.0000
3.1543	5.8309	14000	3.2640	1.0000	1.0000
3.1311	6.2474	15000	3.2485	1.0000	1.0000
3.1059	6.6639	16000	3.2373	1.0000	1.0000
3.1087	7.0804	17000	3.2766	1.0000	1.0000
3.1429	7.4969	18000	3.2375	1.0000	1.0000
3.104	7.9134	19000	3.2643	1.0000	1.0000
3.1288	8.3299	20000	3.2411	1.0000	1.0000
3.1246	8.7464	21000	3.2481	1.0000	1.0000
3.0808	9.1628	22000	3.2583	1.0000	1.0000
3.0996	9.5793	23000	3.2445	1.0000	1.0000
3.0871	9.9958	24000	3.2660	1.0000	1.0000
3.0789	10.4123	25000	3.2566	1.0000	1.0000
3.0828	10.8288	26000	3.2379	1.0000	1.0000
3.1161	11.2453	27000	3.2458	1.0000	1.0000
3.0719	11.6618	28000	3.2473	1.0000	1.0000
3.0838	12.0783	29000	3.2451	1.0000	1.0000
3.0758	12.4948	30000	3.2472	1.0000	1.0000

Framework versions

• Transformers 4.52.3
• Pytorch 2.7.0+cu118
• Datasets 3.6.0
• Tokenizers 0.21.1