metadata
library_name: transformers
license: mit
base_model: distil-whisper/distil-medium.en
tags:
- generated_from_trainer
datasets:
- speech_commands
metrics:
- accuracy
model-index:
- name: distil-medium.en-ft-kws-speech-commands
results:
- task:
name: Audio Classification
type: audio-classification
dataset:
name: Speech Commands
type: speech_commands
config: v0.02
split: test
args: v0.02
metrics:
- name: Accuracy
type: accuracy
value: 0.8066546762589928
distil-medium.en-ft-kws-speech-commands
This model is a fine-tuned version of distil-whisper/distil-medium.en on the Speech Commands dataset. It achieves the following results on the evaluation set:
- Loss: 1.6851
- Accuracy: 0.8067
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: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- total_train_batch_size: 32
- total_eval_batch_size: 32
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.1179 | 1.0 | 1236 | 0.8986 | 0.7990 |
| 0.1177 | 2.0 | 2472 | 0.8863 | 0.8008 |
| 0.0953 | 3.0 | 3708 | 0.9958 | 0.8031 |
| 0.1288 | 4.0 | 4944 | 1.0659 | 0.8017 |
| 0.0575 | 5.0 | 6180 | 1.1709 | 0.8026 |
| 0.0011 | 6.0 | 7416 | 1.1123 | 0.8049 |
| 0.0005 | 7.0 | 8652 | 1.2285 | 0.8049 |
| 0.0006 | 8.0 | 9888 | 1.3904 | 0.8058 |
| 0.001 | 9.0 | 11124 | 1.4603 | 0.8067 |
| 0.0001 | 10.0 | 12360 | 1.6851 | 0.8067 |
Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3