w2v-bert-2.0-slovak-colab-CV17.0

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3823
• Wer: 0.1359

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.1601	1.6393	300	0.4869	0.2985
0.2141	3.2787	600	0.3886	0.2144
0.1323	4.9180	900	0.3180	0.1840
0.0754	6.5574	1200	0.3019	0.1750
0.0401	8.1967	1500	0.3717	0.1525
0.022	9.8361	1800	0.3408	0.1503
0.0083	11.4754	2100	0.3489	0.1413
0.0027	13.1148	2400	0.3681	0.1358
0.0011	14.7541	2700	0.3823	0.1359

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.1
• Datasets 2.19.1
• Tokenizers 0.20.1