w2v-bert-bem-genbed-m-model

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

• Loss: 0.4168
• Wer: 0.5478

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7215	1.1019	200	0.6150	0.7430
0.5519	2.2039	400	0.5605	0.7116
0.4346	3.3058	600	0.4709	0.6378
0.3545	4.4077	800	0.4686	0.5984
0.3004	5.5096	1000	0.4578	0.6203
0.2498	6.6116	1200	0.4245	0.5246
0.23	7.7135	1400	0.4168	0.5478
0.1959	8.8154	1600	0.4212	0.5230
0.1682	9.9174	1800	0.4357	0.5054
0.1459	11.0193	2000	0.4253	0.5296

Framework versions

• Transformers 4.46.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0