base_sami_22k_ftpseudo_ftlabelled_sami_parliament

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 220.5877
• Wer: 0.4061
• Cer: 0.1238

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: 0.0005
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.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: 0.25
• num_epochs: 60.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1059.4936	1.0	446	233.5363	0.4193	0.1316
868.2499	2.0	892	220.7794	0.4036	0.1232
820.7753	3.0	1338	256.3383	0.4162	0.1329
844.8047	4.0	1784	253.8045	0.4216	0.1430
792.2914	5.0	2230	250.6644	0.4473	0.1485
825.7003	6.0	2676	307.4147	0.4676	0.1611
840.0486	7.0	3122	304.3511	0.4777	0.1686

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1
• Datasets 3.2.0
• Tokenizers 0.21.0