mms-1b-all-bemgen-combined-fusion

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2176
• Wer: 0.3968

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: 42
• 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: 5.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
7.7297	0.2538	100	5.6005	0.9996
4.8876	0.5076	200	5.1892	1.1228
4.4438	0.7614	300	4.3481	1.1233
3.7194	1.0152	400	3.3746	1.0120
3.3173	1.2690	500	3.2560	1.0060
3.1776	1.5228	600	2.9594	1.0037
1.1178	1.7766	700	0.2583	0.4834
0.4382	2.0305	800	0.2357	0.4296
0.422	2.2843	900	0.2322	0.4304
0.4101	2.5381	1000	0.2311	0.4307
0.3923	2.7919	1100	0.2303	0.4160
0.382	3.0457	1200	0.2249	0.4061
0.3799	3.2995	1300	0.2240	0.4017
0.3733	3.5533	1400	0.2204	0.3992
0.3757	3.8071	1500	0.2206	0.4031
0.3789	4.0609	1600	0.2192	0.4032
0.3635	4.3147	1700	0.2189	0.4052
0.3583	4.5685	1800	0.2179	0.3989
0.3733	4.8223	1900	0.2176	0.3968

Framework versions

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