aleni/whisper-small-vi

Whisper small vi - Ox

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

• Loss: 0.2559
• Wer: 9.3480

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: Use OptimizerNames.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_steps: 500
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.249	0.0780	1000	0.1739	8.6461
0.2426	0.1559	2000	0.1802	8.6256
0.238	0.2339	3000	0.1851	8.7411
0.2096	0.3119	4000	0.1892	9.4001
0.2076	0.3898	5000	0.1965	8.7169
0.1989	0.4678	6000	0.1952	9.3685
0.2172	0.5458	7000	0.2029	9.0334
0.2145	0.6237	8000	0.2044	9.4430
0.2151	0.7017	9000	0.2079	9.1227
0.2307	0.7797	10000	0.2095	8.9310
0.2065	0.8576	11000	0.2200	9.7595
0.2252	0.9356	12000	0.2177	9.8823
0.1236	1.0136	13000	0.2221	10.3832
0.1242	1.0915	14000	0.2276	9.2549
0.1392	1.1695	15000	0.2272	9.5677
0.1274	1.2475	16000	0.2272	8.6200
0.139	1.3254	17000	0.2301	9.1209
0.1166	1.4034	18000	0.2325	9.1711
0.1507	1.4814	19000	0.2323	9.5472
0.106	1.5593	20000	0.2331	9.6868
0.1264	1.6373	21000	0.2372	8.9440
0.1177	1.7153	22000	0.2394	8.9924
0.1125	1.7932	23000	0.2411	9.1618
0.1272	1.8712	24000	0.2421	9.7669
0.1234	1.9492	25000	0.2441	9.6552
0.091	2.0271	26000	0.2502	8.9272
0.0778	2.1051	27000	0.2499	9.3052
0.0853	2.1831	28000	0.2513	10.2081
0.079	2.2610	29000	0.2532	9.7446
0.0661	2.3390	30000	0.2533	9.7613
0.0782	2.4170	31000	0.2525	9.2884
0.0757	2.4949	32000	0.2550	9.4969
0.0704	2.5729	33000	0.2554	9.4169
0.066	2.6509	34000	0.2560	9.2642
0.0703	2.7288	35000	0.2564	9.2828
0.0707	2.8068	36000	0.2552	9.2977
0.0722	2.8848	37000	0.2566	9.3629
0.081	2.9627	38000	0.2559	9.3480

Framework versions

• Transformers 4.53.0.dev0
• Pytorch 2.8.0.dev20250616+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1