metadata
library_name: transformers
language:
- ne
license: apache-2.0
base_model: openai/whisper-base
tags:
- generated_from_trainer
datasets:
- openslr/openslr
metrics:
- wer
model-index:
- name: Whisper Medium - Kiran Pantha
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: OpenSLR54
type: openslr/openslr
config: default
split: test
args: 'config: ne, split: test'
metrics:
- name: Wer
type: wer
value: 43.58105012370567
Whisper Medium - Kiran Pantha
This model is a fine-tuned version of openai/whisper-base on the OpenSLR54 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2096
- Wer: 43.5811
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-05
- train_batch_size: 8
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.5243 | 0.1200 | 500 | 0.4983 | 77.1832 |
| 0.3685 | 0.2399 | 1000 | 0.3600 | 64.9684 |
| 0.3007 | 0.3599 | 1500 | 0.3094 | 58.0592 |
| 0.2704 | 0.4798 | 2000 | 0.2785 | 54.5038 |
| 0.2529 | 0.5998 | 2500 | 0.2560 | 50.7560 |
| 0.2479 | 0.7198 | 3000 | 0.2407 | 48.4193 |
| 0.2349 | 0.8397 | 3500 | 0.2262 | 46.3850 |
| 0.211 | 0.9597 | 4000 | 0.2170 | 45.2305 |
| 0.1707 | 1.0797 | 4500 | 0.2132 | 44.6990 |
| 0.1506 | 1.1996 | 5000 | 0.2096 | 43.5811 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu121
- Datasets 2.21.0
- Tokenizers 0.19.1