zuazo/whisper-large-v2-pt

Whisper Large-V2 Portuguese

This model is a fine-tuned version of openai/whisper-large-v2 on the mozilla-foundation/common_voice_13_0 pt dataset. It achieves the following results on the evaluation set:

• Loss: 0.4680
• Wer: 5.8752

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-06
• train_batch_size: 32
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0874	3.53	1000	0.1593	4.9765
0.0318	7.05	2000	0.2263	5.4365
0.0121	10.58	3000	0.2966	5.5630
0.005	14.11	4000	0.3400	5.6123
0.0036	17.64	5000	0.3554	5.6600
0.0034	21.16	6000	0.3640	5.6370
0.0021	24.69	7000	0.3714	5.6485
0.0016	28.22	8000	0.3962	5.6255
0.0013	31.75	9000	0.3960	5.6731
0.0009	35.27	10000	0.4107	5.7027
0.0008	38.8	11000	0.3981	5.9869
0.0006	42.33	12000	0.4097	5.7010
0.0005	45.86	13000	0.4226	5.8144
0.0004	49.38	14000	0.4330	5.8259
0.0004	52.91	15000	0.4415	5.7914
0.0003	56.44	16000	0.4490	5.7848
0.0003	59.96	17000	0.4553	5.8013
0.0002	63.49	18000	0.4625	5.7963
0.0002	67.02	19000	0.4663	5.8522
0.0002	70.55	20000	0.4680	5.8752

Framework versions

• Transformers 4.37.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.4
• Tokenizers 0.15.1