metadata

library_name: transformers
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
  - automatic-speech-recognition
  - lozgen
  - mms
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: mms-1b-lozgen-male-model
    results: []

mms-1b-lozgen-male-model

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

Loss: 0.3107
Wer: 0.3264

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
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: 30.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.7665	0.7634	100	3.1610	0.9990
2.6476	1.5267	200	2.2410	0.8953
1.5694	2.2901	300	0.5585	0.7599
0.6761	3.0534	400	0.4566	0.6319
0.5793	3.8168	500	0.4013	0.4822
0.5392	4.5802	600	0.3795	0.4554
0.4809	5.3435	700	0.3730	0.4333
0.4813	6.1069	800	0.3597	0.4230
0.4484	6.8702	900	0.3432	0.3925
0.4418	7.6336	1000	0.3391	0.3947
0.4322	8.3969	1100	0.3339	0.3841
0.3963	9.1603	1200	0.3294	0.3669
0.4104	9.9237	1300	0.3217	0.3635
0.3777	10.6870	1400	0.3177	0.3610
0.3785	11.4504	1500	0.3236	0.3539
0.3682	12.2137	1600	0.3144	0.3468
0.3654	12.9771	1700	0.3122	0.3529
0.3509	13.7405	1800	0.3088	0.3463
0.3412	14.5038	1900	0.3146	0.3347
0.3344	15.2672	2000	0.3108	0.3416
0.3351	16.0305	2100	0.3107	0.3264

Framework versions

Transformers 4.48.0.dev0
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.0