metadata

language:
  - en
license: apache-2.0
base_model: openai/whisper-medium
tags:
  - generated_from_trainer
datasets:
  - QEC
metrics:
  - wer
model-index:
  - name: whisper-medium-quartr
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Quartr Earnings Calls
          type: QEC
          args: 'config: en, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 22.31368880573745

whisper-medium-quartr

This model is a fine-tuned version of openai/whisper-medium on the Quartr Earnings Calls dataset. It achieves the following results on the evaluation set:

Loss: 0.6825
Wer: 22.3137

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: 8.120528078446462e-06
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine_with_restarts
lr_scheduler_warmup_steps: 84
training_steps: 1500
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5817	0.32	100	0.5708	21.9832
0.5817	0.64	200	0.5332	20.1559
0.5253	0.96	300	0.5127	25.4256
0.3177	1.28	400	0.5276	28.5688
0.3603	1.61	500	0.5195	22.2950
0.3374	1.93	600	0.5101	24.3343
0.1734	2.25	700	0.5530	23.1743
0.2002	2.57	800	0.5525	21.1537
0.1894	2.89	900	0.5589	21.7774
0.0868	3.21	1000	0.6291	23.4487
0.0931	3.53	1100	0.6410	21.9208
0.1094	3.85	1200	0.6339	22.5008
0.1007	4.17	1300	0.6698	21.7524
0.0652	4.49	1400	0.6820	22.3262
0.0614	4.82	1500	0.6825	22.3137

Framework versions

Transformers 4.40.0.dev0
Pytorch 2.2.1+cu121
Datasets 2.18.1.dev0
Tokenizers 0.15.2