p4b/whisper-large-v2-lv

Whisper Large-v2 Latvian

This model is a fine-tuned version of p4b/whisper-large-v2-lv on the mozilla-foundation/common_voice_11_0 lv dataset. It achieves the following results on the evaluation set:

• Loss: 0.2593
• Wer: 19.9715

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-07
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 900

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7919	3.03	200	0.2793	22.5806
0.4409	6.05	400	0.2651	20.6072
0.4393	10.01	600	0.2600	20.0664
0.4975	13.04	800	0.2593	19.9715

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 2.0.0.dev20221218+cu116
• Datasets 2.7.1
• Tokenizers 0.13.2