Modelpublic

AkshitSaxena1/whisper-tiny_to_chinese_accent_v3

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face

0likes7downloads

Model Card

Whisper tiny Spanish

This model is a fine-tuned version of openai/whisper-tiny on the Spanish English dataset. It achieves the following results on the evaluation set:

—Loss: 0.3566
—Wer: 14.6942

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-06
—trainbatchsize: 2
—evalbatchsize: 1
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 4
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: linear
—lrschedulerwarmup_steps: 100
—training_steps: 1600

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0338	0.2567	100	0.3285	14.5036
0.0223	0.5135	200	0.3329	14.5671
0.0160	0.7702	300	0.3344	14.5830
0.0185	1.0257	400	0.3374	14.5671
0.0063	1.2824	500	0.3413	14.4083
0.0088	1.5392	600	0.3440	14.6307
0.0114	1.7959	700	0.3442	14.6624
0.0042	2.0513	800	0.3474	14.8372
0.0061	2.3081	900	0.3489	14.7577
0.0041	2.5648	1000	0.3531	14.5830
0.0037	2.8216	1100	0.3527	14.7736
0.0028	3.0770	1200	0.3520	14.7419
0.0034	3.3338	1300	0.3553	14.7419
0.0033	3.5905	1400	0.3557	14.7419
0.0037	3.8472	1500	0.3560	14.7419
0.0026	4.1027	1600	0.3566	14.6942

Framework versions

—Transformers 5.5.0.dev0
—Pytorch 2.10.0+cu128
—Datasets 4.8.4
—Tokenizers 0.22.2