xuliu15/Frisian_32r_LoRA_10mins

LoRA-Frisian-10m

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_6_1 dataset. It achieves the following results on the evaluation set:

• Loss: 1.4867

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	1.0	15	2.3256
1.8865	2.0	30	2.1013
1.8865	3.0	45	1.6160
1.0013	4.0	60	1.5231
0.292	5.0	75	1.5076
0.292	6.0	90	1.4960
0.1192	7.0	105	1.4306
0.1192	8.0	120	1.4704
0.0475	9.0	135	1.4549
0.019	10.0	150	1.4525
0.019	11.0	165	1.4633
0.0074	12.0	180	1.4670
0.0074	13.0	195	1.4709
0.0048	14.0	210	1.4759
0.004	15.0	225	1.4792
0.004	16.0	240	1.4820
0.0035	17.0	255	1.4841
0.0035	18.0	270	1.4855
0.0033	19.0	285	1.4862
0.0032	20.0	300	1.4867

Framework versions

• PEFT 0.10.0
• Transformers 4.40.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1