metadata

license: apache-2.0
tags:
  - generated_from_trainer
base_model: openai/whisper-medium
datasets:
  - generator
metrics:
  - wer
model-index:
  - name: whisper-medium-lug
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: generator
          type: generator
          config: default
          split: train
          args: default
        metrics:
          - type: wer
            value: 61.62227602905569
            name: Wer

whisper-medium-lug

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

Loss: 0.2943
Wer: 61.6223

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 8000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.9437	0.025	200	0.4902	427.9661
0.4586	1.0108	400	0.3298	150.3027
0.3741	1.0357	600	0.3337	143.5835
0.2659	2.0215	800	0.2871	109.6852
0.139	3.0072	1000	0.3437	131.9613
0.1734	3.0322	1200	0.3028	170.8838
0.1072	4.018	1400	0.2943	61.6223
0.0726	5.0038	1600	0.3438	114.7094
0.0751	5.0287	1800	0.3526	73.6683
0.0635	6.0145	2000	0.3629	159.7458
0.0554	7.0003	2200	0.3854	152.1186
0.0549	7.0252	2400	0.3751	98.5472
0.0283	8.011	2600	0.3190	89.2857
0.0349	8.036	2800	0.3452	155.5085
0.0379	9.0218	3000	0.3780	109.7458
0.0316	10.0075	3200	0.3880	101.4528
0.0232	10.0325	3400	0.4144	67.7966
0.0246	11.0183	3600	0.3820	71.0654
0.0192	12.004	3800	0.4022	107.6877
0.0195	12.029	4000	0.4276	126.9976
0.013	13.0147	4200	0.4128	115.3753
0.0154	14.0005	4400	0.4371	126.6949
0.0109	14.0255	4600	0.4213	142.2518
0.0133	15.0113	4800	0.4075	170.1574
0.011	15.0363	5000	0.4454	116.1622
0.0104	16.022	5200	0.3950	79.5400
0.0079	17.0078	5400	0.4330	109.2010
0.0083	17.0328	5600	0.4308	137.5303
0.0064	18.0185	5800	0.4178	96.2470
0.0057	19.0042	6000	0.4104	99.7579
0.0076	19.0293	6200	0.4132	117.0702
0.0062	20.015	6400	0.4404	146.2470
0.0035	21.0008	6600	0.4488	128.4504
0.0045	21.0257	6800	0.4415	91.0412
0.0043	22.0115	7000	0.4477	89.5884
0.0038	22.0365	7200	0.4550	82.5666
0.0028	23.0222	7400	0.4451	77.4213
0.003	24.008	7600	0.4424	78.5109
0.0033	24.033	7800	0.4448	73.4867
0.0041	25.0188	8000	0.4455	86.4407

Framework versions

Transformers 4.41.0.dev0
Pytorch 2.2.0
Datasets 2.16.1
Tokenizers 0.19.1