w2v-bert-2.0-CV_Fleurs_AMMI_ALFFA-sw-10hrs-v1

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6790
• Wer: 0.2525
• Cer: 0.0897

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: 4
• 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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2.9209	1.0	710	0.9107	0.4690	0.1399
0.6799	2.0	1420	0.6451	0.3166	0.1020
0.5118	3.0	2130	0.6325	0.2602	0.0900
0.4435	4.0	2840	0.5829	0.2610	0.0951
0.3857	5.0	3550	0.5528	0.2585	0.0952
0.3363	6.0	4260	0.5604	0.2449	0.0863
0.3312	7.0	4970	0.6122	0.3529	0.1307
0.3306	8.0	5680	0.5529	0.2572	0.0931
0.2915	9.0	6390	0.6499	0.2584	0.0929
0.2828	10.0	7100	0.6233	0.2678	0.0954
0.2664	11.0	7810	0.6266	0.2567	0.0904
0.2473	12.0	8520	0.6285	0.2561	0.0894
0.2289	13.0	9230	0.6137	0.2531	0.0901
0.2102	14.0	9940	0.6440	0.2483	0.0891
0.1976	15.0	10650	0.7161	0.2724	0.0957
0.1971	16.0	11360	0.6790	0.2525	0.0897

Framework versions

• Transformers 4.46.0
• Pytorch 2.1.0+cu118
• Datasets 3.0.2
• Tokenizers 0.20.1