mms-1b-all-bemgen-combined-m50f50-dnn-62-0.02

This model is a fine-tuned version of facebook/mms-1b-all on the BEMGEN - DEB dataset. It achieves the following results on the evaluation set:

• Loss: 14.2635
• Cer: 6.3580

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: 4
• seed: 62
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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_steps: 100
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
13.7670	0.8439	100	14.2632	6.3579
16.7331	1.6835	200	14.2633	6.3563
14.1971	2.5232	300	14.2633	6.3589
14.7039	3.3629	400	14.2632	6.3576
12.2117	4.2025	500	14.2632	6.3592
14.8243	5.0422	600	14.2633	6.3590
16.0667	5.8861	700	14.2634	6.3591
12.5454	6.7257	800	14.2632	6.3577
15.6335	7.5654	900	14.2635	6.3573
13.1267	8.4051	1000	14.2633	6.3592
13.8439	9.2447	1100	14.2632	6.3579

Framework versions

• Transformers 4.53.0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.4