whisper-large-v3-turbo-greek-greece

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 0.3006
• Model Preparation Time: 0.0067
• Wer Ortho: 26.8701
• Wer: 11.0591

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Wer Ortho	Wer
0.6129	0.0204	32	0.2996	0.0067	29.5644	14.1172
0.5021	0.0408	64	0.3064	0.0067	29.0414	14.0381
0.5068	0.0612	96	0.3116	0.0067	29.3261	14.2226
0.5049	0.0816	128	0.3000	0.0067	28.5317	13.8404
0.5289	0.1020	160	0.3261	0.0067	29.5048	14.3413
0.4464	0.1224	192	0.3336	0.0067	29.5115	14.2292
0.4324	0.1428	224	0.3267	0.0067	28.9157	13.8865
0.4376	0.1632	256	0.3319	0.0067	29.8292	14.7367
0.4623	0.1836	288	0.3493	0.0067	29.5379	14.2424
0.4494	0.2040	320	0.3210	0.0067	28.9620	13.7415
0.4541	0.2244	352	0.3284	0.0067	29.0348	13.7349
0.4262	0.2448	384	0.3169	0.0067	28.7369	13.2077
0.4365	0.2652	416	0.3131	0.0067	28.7369	13.1484
0.4701	0.2856	448	0.3144	0.0067	28.5118	13.0759
0.4256	0.3060	480	0.3017	0.0067	28.4126	13.2868
0.4437	0.3264	512	0.3080	0.0067	28.6111	13.3724
0.4147	0.3468	544	0.3141	0.0067	28.1875	12.4629
0.4089	0.3672	576	0.3150	0.0067	28.2669	12.9638
0.4448	0.3876	608	0.3221	0.0067	28.5979	12.8913
0.4354	0.4080	640	0.3091	0.0067	28.1875	12.7859
0.3877	0.4284	672	0.3174	0.0067	28.0816	12.5816
0.4411	0.4488	704	0.3130	0.0067	28.4788	13.0693
0.4056	0.4692	736	0.3168	0.0067	28.3000	12.4366
0.4043	0.4896	768	0.3291	0.0067	28.1213	12.4695
0.4265	0.5100	800	0.3230	0.0067	27.8565	12.0082
0.4075	0.5304	832	0.3062	0.0067	27.4990	12.1400
0.4075	0.5508	864	0.3137	0.0067	27.6711	12.0675
0.4312	0.5712	896	0.3086	0.0067	27.6777	11.8302
0.4294	0.5916	928	0.3034	0.0067	27.4262	12.1202
0.4268	0.6120	960	0.3126	0.0067	27.9558	12.3707
0.4385	0.6325	992	0.3156	0.0067	28.0087	12.3113
0.3886	0.6529	1024	0.3043	0.0067	27.6115	11.8566
0.3761	0.6733	1056	0.3076	0.0067	27.2541	11.8434
0.4109	0.6937	1088	0.3125	0.0067	27.4924	12.0214
0.4231	0.7141	1120	0.3025	0.0067	27.3070	11.8368
0.4126	0.7345	1152	0.3058	0.0067	27.3070	11.7577
0.3999	0.7549	1184	0.3056	0.0067	26.9694	11.4677
0.4354	0.7753	1216	0.3031	0.0067	27.4725	11.6589
0.4036	0.7957	1248	0.2979	0.0067	27.2210	11.4611
0.3972	0.8161	1280	0.3043	0.0067	27.3467	11.4348
0.3478	0.8365	1312	0.3055	0.0067	27.4063	11.5139
0.3915	0.8569	1344	0.3015	0.0067	27.1813	11.4743
0.4092	0.8773	1376	0.2999	0.0067	27.0356	11.3952
0.4042	0.8977	1408	0.2974	0.0067	27.0025	11.3227
0.4332	0.9181	1440	0.3026	0.0067	27.2739	11.4282
0.4164	0.9385	1472	0.2996	0.0067	27.0158	11.1777
0.3774	0.9589	1504	0.3009	0.0067	27.0820	11.0987
0.3934	0.9793	1536	0.3006	0.0067	26.8701	11.0591
0.3695	0.9997	1568	0.3020	0.0067	26.9032	11.0921

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1
• Datasets 3.6.0
• Tokenizers 0.21.1