nurzhanit/whisper-enhanced-ml

Whisper Small Hi - Sanchit Gandhi

This model is a fine-tuned version of nurzhanit/whisper-enhanced-ml on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0000
• Wer: 0.0

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: 1e-05
• 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: 200
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0	16.6667	100	0.0000	0.0
0.0	33.3333	200	0.0000	0.0
0.0	50.0	300	0.0000	0.0
0.0	66.6667	400	0.0000	0.0
0.0	83.3333	500	0.0000	0.0
0.0	100.0	600	0.0000	0.0
0.0	116.6667	700	0.0000	0.0
0.0	133.3333	800	0.0000	0.0
0.0	150.0	900	0.0000	0.0
0.0	166.6667	1000	0.0000	0.0
0.0	183.3333	1100	0.0000	0.0
0.0	200.0	1200	0.0000	0.0
0.0	216.6667	1300	0.0000	0.0
0.0	233.3333	1400	0.0000	0.0
0.0	250.0	1500	0.0000	0.0
0.0	266.6667	1600	0.0000	0.0
0.0	283.3333	1700	0.0000	0.0
0.0	300.0	1800	0.0000	0.0
0.0	316.6667	1900	0.0000	0.0
0.0	333.3333	2000	0.0000	0.0
0.0	350.0	2100	0.0000	0.0
0.0	366.6667	2200	0.0000	0.0
0.0	383.3333	2300	0.0000	0.0
0.0	400.0	2400	0.0000	0.0
0.0	416.6667	2500	0.0000	0.0
0.0	433.3333	2600	0.0000	0.0
0.0	450.0	2700	0.0000	0.0
0.0	466.6667	2800	0.0000	0.0
0.0	483.3333	2900	0.0000	0.0
0.0	500.0	3000	0.0000	0.0
0.0	516.6667	3100	0.0000	0.0
0.0	533.3333	3200	0.0000	0.0
0.0	550.0	3300	0.0000	0.0
0.0	566.6667	3400	0.0000	0.0
0.0	583.3333	3500	0.0000	0.0
0.0	600.0	3600	0.0000	0.0
0.0	616.6667	3700	0.0000	0.0
0.0	633.3333	3800	0.0000	0.0
0.0	650.0	3900	0.0000	0.0
0.0	666.6667	4000	0.0000	0.0

Framework versions

• Transformers 4.40.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.2
• Tokenizers 0.19.1