mms-1b-all-bigc-combined-25hrs-62

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

• Loss: 1.3401
• Wer: 0.7082

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	Wer
7.5381	0.1300	100	1.7495	0.9106
0.8251	0.2601	200	1.5294	0.7411
0.8076	0.3901	300	1.4162	0.7213
0.7584	0.5202	400	1.3402	0.7082
0.7029	0.6502	500	1.4399	0.7060
0.7099	0.7802	600	1.3644	0.6966
0.7454	0.9103	700	1.3533	0.6897

Framework versions

• Transformers 4.53.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.0