./whisper-large-cit-synth-do02-wd0-lr1e-06-200

This model is a fine-tuned version of openai/whisper-large-v3 on the SF 200 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4683
• Wer: 23.3684

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: 4
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 300
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.9421	4.4444	50	0.6841	27.5789
0.5077	8.8889	100	0.4324	24.8421
0.2787	13.3333	150	0.4265	23.3684
0.177	17.7778	200	0.4451	22.7368
0.1312	22.2222	250	0.4609	22.5263
0.1129	26.6667	300	0.4683	23.3684

Framework versions

• Transformers 4.42.3
• Pytorch 1.13.1+cu117
• Datasets 2.20.0
• Tokenizers 0.19.1