NbAiLab/whisper-tiny-nob

Whisper Tiny Norwegian Bokmål

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

• Loss: 0.5100
• Wer: 24.8782

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: 3e-06
• train_batch_size: 256
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 1000
• training_steps: 100000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.8819	0.01	1000	1.1869	61.9671
1.6425	0.02	2000	0.9991	53.6541
1.548	0.03	3000	0.9147	50.2132
1.4636	0.04	4000	0.8605	47.0767
1.4113	0.05	5000	0.8253	45.7369
1.3484	0.01	6000	0.7946	43.4531
1.3127	0.02	7000	0.7740	42.2655
1.2994	0.03	8000	0.7551	40.8952
1.265	0.04	9000	0.7378	39.8599
1.2458	0.05	10000	0.7257	39.8904
1.2257	0.06	11000	0.7114	39.7990
1.2126	0.07	12000	0.6972	37.8806
1.1971	0.08	13000	0.6871	37.3021
1.1786	1.01	14000	0.6786	37.4239
1.1486	1.02	15000	0.6703	36.9976
1.1505	1.03	16000	0.6647	36.3581
1.1238	1.04	17000	0.6559	36.3886
1.1184	1.05	18000	0.6509	36.5104
1.115	1.06	19000	0.6452	35.9927
1.1013	1.07	20000	0.6382	34.5006
1.0969	1.08	21000	0.6331	34.3484
1.0784	2.0	22000	0.6304	34.2875
1.0774	2.01	23000	0.6249	34.1048
1.0719	2.02	24000	0.6194	33.8307
1.0638	2.03	25000	0.6158	32.9781
1.0592	2.04	26000	0.6105	32.6431
1.0493	2.05	27000	0.6041	32.7345
1.047	2.06	28000	0.6040	32.7649
1.0323	2.07	29000	0.5984	31.6078
1.0189	3.0	30000	0.5957	31.3033
1.0078	3.01	31000	0.5924	31.4251
1.0146	3.02	32000	0.5940	31.3033
1.0128	3.03	33000	0.5892	31.0292
1.0025	3.04	34000	0.5873	31.1815
0.999	3.05	35000	0.5838	30.6334
1.0045	3.06	36000	0.5799	30.4202
1.0005	3.07	37000	0.5770	30.1766
1.0017	3.08	38000	0.5733	29.6590
0.9878	4.01	39000	0.5745	30.2680
0.9854	4.02	40000	0.5720	30.0548
0.9624	4.03	41000	0.5703	29.5981
0.9639	4.04	42000	0.5681	29.5067
0.9569	4.05	43000	0.5679	29.6285
0.9682	4.06	44000	0.5643	29.5676
0.9539	4.07	45000	0.5601	29.5676
0.946	4.08	46000	0.5562	29.7199
0.9429	5.01	47000	0.5592	29.2935
0.9462	5.02	48000	0.5540	29.0804
0.9312	5.03	49000	0.5535	29.2935
0.9462	5.04	50000	0.5536	28.6845
0.922	5.05	51000	0.5539	28.7150
0.9253	5.06	52000	0.5510	28.8368
0.9065	0.01	53000	0.5493	28.5932
0.9096	0.02	54000	0.5490	28.5018
0.9329	0.03	55000	0.5483	28.2887
0.9181	0.04	56000	0.5471	27.9842
0.914	0.05	57000	0.5457	28.4105
0.9149	0.06	58000	0.5449	27.5883
0.9092	0.07	59000	0.5405	27.8319
0.9101	0.08	60000	0.5402	27.3447
0.9046	1.01	61000	0.5374	27.5579
0.8917	1.02	62000	0.5390	27.7406
0.8993	1.03	63000	0.5386	27.4056
0.8875	1.04	64000	0.5361	26.8575
0.8892	1.05	65000	0.5358	27.3447
0.8929	1.06	66000	0.5346	26.7357
0.8703	0.01	67000	0.5332	26.8270
0.8709	0.02	68000	0.5336	26.7052
0.8917	0.03	69000	0.5329	27.0706
0.8867	0.04	70000	0.5323	26.3398
0.8778	0.05	71000	0.5315	27.2838
0.8757	0.06	72000	0.5317	26.2485
0.8726	0.07	73000	0.5269	26.6443
0.8792	0.08	74000	0.5268	26.1571
0.8706	1.01	75000	0.5247	26.1571
0.8585	1.02	76000	0.5265	26.3703
0.8659	1.03	77000	0.5262	26.7357
0.8551	1.04	78000	0.5249	26.0658
0.8572	1.05	79000	0.5249	26.2789
0.8612	1.06	80000	0.5235	25.7613
0.8598	1.07	81000	0.5208	25.7004
0.8686	1.08	82000	0.5214	25.7004
0.8503	2.0	83000	0.5214	25.7004
0.8545	2.01	84000	0.5215	28.2278
0.8594	2.02	85000	0.5186	25.6699
0.86	2.03	86000	0.5196	25.5786
0.8514	2.04	87000	0.5203	25.1827
0.8505	2.05	88000	0.5164	28.0146
0.8512	2.06	89000	0.5174	25.0914
0.8495	2.07	90000	0.5141	25.5481
0.8381	3.0	91000	0.5130	24.9695
0.8253	3.01	92000	0.5147	25.5786
0.8387	3.02	93000	0.5168	24.9086
0.8425	3.03	94000	0.5135	25.2436
0.8339	3.04	95000	0.5162	25.6699
0.8402	3.05	96000	0.5147	25.7308
0.8396	3.06	97000	0.5143	25.6699
0.8432	3.07	98000	0.5100	24.8782
0.844	3.08	99000	0.5100	25.0609
0.8333	4.01	100000	0.5128	24.9695

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2