You can deploy this model and start earning money today!

KasuleTrevor/
cdli-whisper-lg-sulaimank-full-a40-5e-5-cosine

Name: cdli-whisper-lg-sulaimank-full-a40-5e-5-cosine
Availability: InStock

transformers

safetensors

whisper

automatic-speech-recognition

generated_from_trainer

cdli/ugandan_luganda_nonstandard_speech_v1.0

sulaimank/whisper-small-luganda-400hrs

finetune

apache-2.0

endpoints_compatible

Views

No views yet

cdli-whisper-lg-sulaimank-full-a40-5e-5-cosine

This model is a fine-tuned version of sulaimank/whisper-small-luganda-400hrs on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.2813
Wer: 0.6317
Cer: 0.3505
test_cer = 0.2537
test_wer = 0.5058

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: 5e-05
train_batch_size: 8
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: cosine
lr_scheduler_warmup_steps: 500
training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.3683	0.6784	500	1.5102	0.6869	0.3945
0.9356	1.3569	1000	1.2470	0.6488	0.3617
0.7626	2.0353	1500	1.1881	0.6305	0.3436
0.8868	2.7137	2000	1.1537	0.6429	0.3601
0.6717	3.3921	2500	1.1653	0.6398	0.3553
0.6589	4.0706	3000	1.1825	0.6339	0.3466
0.65	4.7490	3500	1.1822	0.6326	0.3519
0.5557	5.4274	4000	1.2099	0.6307	0.3465
0.5777	6.1058	4500	1.2217	0.6278	0.3512
0.4821	6.7843	5000	1.2478	0.6339	0.3524
0.4833	7.4627	5500	1.2609	0.6293	0.3498
0.5251	8.1411	6000	1.2689	0.6322	0.3501
0.5061	8.8195	6500	1.2734	0.6329	0.3511
0.5475	9.4980	7000	1.2812	0.6341	0.3537
0.428	10.1764	7500	1.2793	0.6341	0.3520
0.4676	10.8548	8000	1.2812	0.6334	0.3528
0.5068	11.5332	8500	1.2813	0.6324	0.3519
0.5324	12.2117	9000	1.2813	0.6333	0.3515
0.5551	12.8901	9500	1.2812	0.6330	0.3523
0.4697	13.5685	10000	1.2813	0.6317	0.3505

Framework versions

Transformers 4.57.6
Pytorch 2.7.1+cu118
Datasets 3.6.0
Tokenizers 0.22.2

KasuleTrevor/cdli-whisper-lg-sulaimank-full-a40-5e-5-cosine

cdli-whisper-lg-sulaimank-full-a40-5e-5-cosine

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

KasuleTrevor/
cdli-whisper-lg-sulaimank-full-a40-5e-5-cosine