FULL6

This model is a fine-tuned version of openai/whisper-large-v3 on the 7263 FULL-2024-11-28 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4058
  • Wer Ortho: 22.3744
  • Wer: 15.8352

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: 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: 300
  • training_steps: 1400
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Ortho Wer
0.6852 0.4896 200 0.4802 27.7318 21.0033
0.5153 0.9792 400 0.4355 24.3891 17.7718
0.4247 1.4688 600 0.4202 23.1486 16.5923
0.4106 1.9584 800 0.4087 22.8364 16.4130
0.3575 2.4480 1000 0.4097 22.6283 16.0902
0.3377 2.9376 1200 0.4045 23.1237 16.6281
0.3157 3.4272 1400 0.4058 22.3744 15.8352

Framework versions

  • Transformers 4.45.1
  • Pytorch 1.13.1+cu117
  • Datasets 3.0.1
  • Tokenizers 0.20.0
Downloads last month
20
Safetensors
Model size
1.61B params
Tensor type
FP16
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for Makkoen/whisper-large-v3-cit-do005-wd0-lr3e-06-steps1400-FULL5b

Finetuned
(350)
this model