metadata

library_name: peft
language:
  - it
base_model: b-brave/asr_double_training_15-10-2024_merged
tags:
  - generated_from_trainer
datasets:
  - ASR_BB_and_EC
metrics:
  - wer
model-index:
  - name: Whisper Medium
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: ASR_BB_and_EC
          type: ASR_BB_and_EC
          config: default
          split: test
          args: default
        metrics:
          - type: wer
            value: 35.93556381660471
            name: Wer

Whisper Medium

This model is a fine-tuned version of b-brave/asr_double_training_15-10-2024_merged on the ASR_BB_and_EC dataset. It achieves the following results on the evaluation set:

Loss: 0.4742
Wer: 35.9356

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: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: reduce_lr_on_plateau
lr_scheduler_warmup_steps: 100
num_epochs: 6
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7705	0.8929	100	0.4884	36.5551
0.7194	1.7857	200	0.4838	36.6791
0.7372	2.6786	300	0.4806	36.4312
0.6967	3.5714	400	0.4782	36.3073
0.649	4.4643	500	0.4761	36.3073
0.7127	5.3571	600	0.4742	35.9356

Framework versions

PEFT 0.13.2
Transformers 4.45.2
Pytorch 2.2.0
Datasets 3.1.0
Tokenizers 0.20.3