wav2vec2-large-mms-1b-amharic-colab

This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6247
• Wer: 0.5047

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: 0.001
• 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_steps: 100
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
15.6099	1.1364	50	3.3812	0.9995
1.174	2.2727	100	0.6846	0.5174
0.6566	3.4091	150	0.6247	0.5047

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0