wav2vec2-large-mms-1b-aft-el-CY

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

• Loss: 0.6250
• Wer: 0.4330

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: 2
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.573	0.2584	100	1.0665	0.6766
0.9412	0.5168	200	0.8605	0.5970
0.7877	0.7752	300	0.7948	0.5750
1.0771	1.0336	400	0.7438	0.5479
0.718	1.2920	500	0.7305	0.5346
0.6581	1.5504	600	0.7471	0.5057
0.9717	1.8088	700	0.7074	0.4965
0.6783	2.0672	800	0.6970	0.5007
0.6062	2.3256	900	0.6747	0.4731
0.8629	2.5840	1000	0.6719	0.4912
0.6127	2.8424	1100	0.6501	0.4662
0.5476	3.1008	1200	0.6517	0.4508
0.5371	3.3592	1300	0.6532	0.4826
0.5478	3.6176	1400	0.6404	0.4498
0.5208	3.8760	1500	0.6335	0.4469
0.7972	4.1344	1600	0.6376	0.4379
0.4907	4.3928	1700	0.6365	0.4362
0.4927	4.6512	1800	0.6320	0.4330
0.8188	4.9096	1900	0.6250	0.4330

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.1.1
• Tokenizers 0.22.1