mms-1b-mdc-rutoro-asr-1.0.0

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

• Loss: 0.2358
• Wer: 0.2124
• Cer: 0.0430

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• 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_ratio: 0.1
• num_epochs: 50.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
7.4756	1.0	84	2.9109	1.0	1.0
2.8944	2.0	168	2.8200	1.0	1.0
2.1261	3.0	252	0.6012	0.7222	0.1484
0.8654	4.0	336	0.3079	0.3961	0.0738
0.6392	5.0	420	0.2751	0.3314	0.0629
0.554	6.0	504	0.2317	0.2795	0.0530
0.4377	7.0	588	0.2330	0.2648	0.0507
0.4423	8.0	672	0.2113	0.2423	0.0475
0.2591	9.0	756	0.2048	0.2332	0.0448
0.4142	10.0	840	0.2072	0.2284	0.0445
0.341	11.0	924	0.2143	0.2292	0.0449
0.266	12.0	1008	0.2041	0.2204	0.0429
0.2572	13.0	1092	0.1940	0.2185	0.0428
0.1793	14.0	1176	0.2166	0.2209	0.0436
0.331	15.0	1260	0.2247	0.2167	0.0427
0.2035	16.0	1344	0.2147	0.2143	0.0429
0.2905	17.0	1428	0.2189	0.2128	0.0420
0.3872	18.0	1512	0.2185	0.2149	0.0427
0.256	19.0	1596	0.2276	0.2140	0.0420
0.1576	20.0	1680	0.2336	0.2129	0.0424
0.1607	21.0	1764	0.2339	0.2128	0.0426
0.148	22.0	1848	0.2358	0.2124	0.0430

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.0+cu128
• Datasets 4.4.0
• Tokenizers 0.22.1