w2v-bert-2.0-tigre-colab-CV17.0-v2

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: 1.6193
• Wer: 0.4317

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_ratio: 0.15
• training_steps: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.7881	13.7931	200	0.9737	0.6175
0.1599	27.5862	400	1.2407	0.5310
0.0256	41.3793	600	1.3566	0.4781
0.0036	55.1724	800	1.5251	0.4554
0.0058	68.9655	1000	1.4813	0.4699
0.0023	82.7586	1200	1.5533	0.4435
0.0001	96.5517	1400	1.5861	0.4372
0.0001	110.3448	1600	1.6056	0.4362
0.0001	124.1379	1800	1.6159	0.4326
0.0001	137.9310	2000	1.6193	0.4317

Framework versions

• Transformers 4.42.4
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1