Thecoder3281f/whisper-small-hi-nscsmall-1000

Whisper Small NSC small (1000 steps) - Jarrett Er

This model is a fine-tuned version of Thecoder3281f/whisper-small-hi-commonvoice17-1000 on the NSC Small section dataset. It achieves the following results on the evaluation set:

• Loss: 0.0676
• Wer: 3.1233

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: 0.0001
• train_batch_size: 16
• 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: 100
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0806	0.2941	100	0.0737	3.4549
0.0618	0.5882	200	0.0690	3.2062
0.0689	0.8824	300	0.0655	3.0265
0.0385	1.1765	400	0.0652	3.1509
0.0441	1.4706	500	0.0653	3.1647
0.0389	1.7647	600	0.0652	3.0404
0.032	2.0588	700	0.0646	3.1786
0.0264	2.3529	800	0.0672	3.1095
0.0307	2.6471	900	0.0672	3.1647
0.0266	2.9412	1000	0.0676	3.1233

Framework versions

• PEFT 0.14.0
• Transformers 4.45.2
• Pytorch 2.5.1+cu124
• Datasets 3.2.1.dev0
• Tokenizers 0.20.3