mms_sesame_knn_vc

This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9116
• Wer: 0.5995
• Cer: 0.3471
• Bertscore Precision: 0.6224
• Bertscore Recall: 0.6664
• Bertscore F1: 0.6400

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use 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: 50
• training_steps: 2000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Bertscore Precision	Bertscore Recall	Bertscore F1
1.9216	21.0085	500	1.6593	0.5909	0.3361	0.6275	0.6691	0.6440
1.7172	43.0055	1000	1.7099	0.6058	0.3503	0.6315	0.6703	0.6469
1.302	65.0025	1500	1.7937	0.5971	0.3442	0.6320	0.6731	0.6484
1.0859	86.011	2000	1.9116	0.5995	0.3471	0.6224	0.6664	0.6400

Framework versions

• Transformers 4.57.6
• Pytorch 2.8.0+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2