mms-300m-librispeech-adapter-model

This model is a fine-tuned version of facebook/mms-300m on the LIBRI10H - ENG dataset. It achieves the following results on the evaluation set:

• Loss: 2.8895
• Wer: 1.0

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use 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: 2.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.0414	0.2899	100	2.9197	1.0
2.8561	0.5797	200	2.8877	1.0
2.8552	0.8696	300	2.8849	1.0
2.842	1.1594	400	2.9165	1.0
2.8444	1.4493	500	2.8822	1.0
2.8381	1.7391	600	2.8861	1.0

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0